Annex 10 PDF Page Vanderlaan, ASM, Taggart, CT, Serdynska, AR

Fisheries and Oceans Canada (DFO) Annex 10

PDF Page

Vanderlaan, A.S.M., Taggart, C.T., Serdynska, A.R., Kenney, R.D., and Brown, M.W. 2008. 2 Reducing the risk of lethal encounters: vessels and right whales in the Bay of Fundy and on the Scotian Shelf. Endang. Spec. Res. 4:283–297.

Veirs, S., Veirs, V., and Wood, J.D. 2016. Ship noise extends to frequencies used for 17 echolocation by endangered killer whales. PeerJ, 4, p.e1657.

Yang, Z., Hollebone, B.P., Zhang, G., Brown, C.E., Yang, C., Lambert, P., Wang, Z., 53 Landriault, M., and Shah, K. 2017. Fate of Photodegraded Diluted Bitumen in Seawater, Proceedings of the 2017 International Oil Spill Conference, American Petroleum Institute, Washington, D.C., pp. 2286-2305.

Yang, Z., Zhang, G., Hollebone, B.P., Brown, C.E., Yang, C., Lambert, P., Landriault, M., 73 and Shah, K. 2017. Fate of Oxygenated Components for Solar Irradiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar on Environmental Contamination and Response, pp. 415-440, Environment and Climate Change Canada, Ottawa, ON.

Yang, Z., Zhang, G., Hollebone, B.P., Brown, C.E., Yang, C., Lambert, P., Wang, Z., 99 Landriault, M., and Shah, K. 2017. Fate of Oxygenated Components for Solar Irradiated Diluted Bitumen Mixed with Seawater, Environmental Pollution, Vol. 231, pp. 622-634, http://dx.doi.org/10.1016/j.envpol.2017.08.043.

Yergeau, E., Maynard, C., Sanschagrin, S., Champagne, J., Juck, D., Lee, K., and Greer, C. 112 2015. Microbial Community Composition, Functions, and Activities in the Gulf of Mexico 1 Year after the Deepwater Horizon Accident. Applied and Environmental Microbiology. 81: 5855-5866.

Yeung, C.W., Lee, K., Cobanli, S., King, T., Bugden, J., Whyte, L., and Greer, C. 2015. 148 Characterization of the microbial community structure and the physicochemical properties of produced water and seawater from the Hibernia oil production platform. Environmental Science Pollution Research International, 22, 17697–715.

Zhao, L., Torlapati, J., King, T., Robinson, B., Boufadel, M., and Lee, K. 2014. A numerical 168 model to simulate the droplet formation process resulting from the release of diluted bitumen products in marine environment. International Oil Spill Conference Proceedings: May 2014, Vol. 2014, 1:449-462.

001

Vol. 4: 283–297, 2008 ENDANGERED SPECIES RESEARCH Printed June 2008 doi: 10.3354/esr00083 Endang Species Res Published online April 16, 2008

OPEN ACCESS Reducing the risk of lethal encounters: vessels and right whales in the Bay of Fundy and on the Scotian Shelf

Angelia S. M. Vanderlaan1,*, Christopher T. Taggart1, Anna R. Serdynska1, 2 3, 4 Robert D. Kenney , Moira W. Brown

1Department of Oceanography, Dalhousie University, Halifax, Nova Scotia B3H 4J1, Canada 2Graduate School of Oceanography, University of Rhode Island, Narragansett, Rhode Island 02882, USA 3New England Aquarium, Boston, Massachusetts 02110, USA 4Canadian Whale Institute, Box 633, Bolton, Ontario L7E 5T4, Canada

ABSTRACT: The North Atlantic right whale Eubalaena glacialis is endangered, in part, due to vessel- strike mortality. We use vessel traffic and right whale survey data (~3 nautical miles [n miles], ~5.6 km resolution) for the Bay of Fundy and on the Scotian Shelf (northwest Atlantic) to determine the relative risk of lethal vessel encounters by using 2 estimates: (1) the event — the relative probability of a vessel encountering a right whale, and (2) the consequence — the probability of a lethal injury arising from an encounter. For the Bay of Fundy region our estimates demonstrate that the relative risk of lethal collision could be reduced by 62% by means of an amendment to the traffic separation scheme (TSS) that intersects a Right Whale Conservation Area. In the Roseway Basin region of the Scotian Shelf, the majority of vessels navigate outside of a Right Whale Conservation Area, although the highest relative risk is concentrated within the Conservation Area where fewer vessels navigate at greater speed. Here, our estimates demonstrate that a seasonal recommendatory area to be avoided (ATBA) could be designed to reduce the risk imposed by vessels upon right whales in the region. Our estimates contributed to the International Maritime Organisation (IMO) adoption of a TSS amendment in the Bay of Fundy and an ATBA on the Scotian Shelf. Thus, the goal of achieving the greatest reduction in the risk of lethal vessel-encounters with whales, balanced by some minimal disruption to vessel operations while maintaining safe navigation, can be achieved.

KEY WORDS: Right whale · Eubalaena glacialis · Mortality · Vessel · Shipping · Strike · Collision · Lethal encounter · Fundy · Roseway

Resale or republication not permitted without written consent of the publisher

INTRODUCTION (1999) estimated extinction probabilities centered on the year 2200 based on contemporary population The North Atlantic right whale Eubalaena glacialis dynamics. Hypotheses related to species recovery lim- (Rosenbaum et al. 2000), hereafter referred to as right itations include those associated with reproductive rate whale, is considered one of the most endangered large (Knowlton et al. 1994, Kraus et al. 2001), genetic vari- whales (Caswell et al. 1999, Kraus et al. 2005) with ability (Waldick et al. 2002), prey-field dynamics (Ken- uncertain population estimates around 300 (±10%) ney et al. 2001, Baumgartner et al. 2007, Michaud & and 350 ind. in the North Atlantic (IWC 2001, Kraus & Taggart 2007), and deleterious human activity (Kraus Rolland 2007). Recent analyses revealed a marginally 1990, Knowlton & Kraus 2001, Kraus et al. 2005). With increasing growth rate of 1.03 in 1980 that syste- approximately one-half of the deaths reported being matically declined to a marginally decreasing rate of caused by human activities (Moore et al. 2007) and 0.98 by 1995 (Fujiwara & Caswell 2001). Caswell et al. vessel strikes accounting for 53% of the determined

*Email: [email protected] © Inter-Research 2008 · www.int-res.com 002

284 Endang Species Res 4: 283–297, 2008

deaths in necroposied whales (Campbell-Malone et al. The 2 most simple and practical methods of decreas- 2008), we focus our study on the probability of lethal ing the likelihood of a vessel strike to a whale are vessel collisions. altering vessel traffic routing in and around known The recovery of the right whale is, in part, contingent whale habitats (to decrease the probability of whale on a reduction in the number of lethal vessel-strikes encounter) or reducing vessel speeds (to decrease the (e.g. Caswell et al. 1999, Fujiwara & Caswell 2001, IWC probability of a lethal injury in the case of an 2001, Kraus et al. 2005). As the right whale appears on a encounter). Only the vessel re-routing option will per capita basis to be more prone to vessel strikes than reduce the concurrence, both spatially and temporally, all other large whales (Vanderlaan & Taggart 2007), of vessels and whales. Only the reduced vessel-speed changes to ocean-going vessel operations must be im- option will decrease the likelihood of a lethal injury plemented to protect the species (Kraus et al. 2005), should an encounter occur (cf. Vanderlaan & Taggart particularly in coastal and shelf regions. The 3 primary 2007). In combining the bases of the above 2 options a means of reducing the likelihood of vessels striking decreased risk (decreased probability of event and right whales include the education of mariners, techno- decreased consequence, i.e. lethality, if the event logical methodologies for detecting whales and warn- occurs) of a lethal collision between a vessel and a ing mariners of whales and warning whales of vessels, whale accrues (Fig. 1). and changing typical vessel operations through altered Right whales are migratory animals and a large pro- traffic routing and vessel speed restrictions (Knowlton portion of the population occupies 2 primary feeding & Brown 2007). habitats in the waters of Atlantic Canada during June Technological methods for alerting mariners to the

1 1 presence of right whales include marine communica- a tion relays of whale sightings by and to vessels transit- 0.9 0.9

ing right whale habitat (Brown et al. 2007). Passive whale 0.8 0.8 encounter a acoustics can be used to monitor and geo-locate right 0.7 0.7 whales (Matthews et al. 2001, Laurinolli et al. 2003, Vanderlaan et al. 2003, Mellinger et al. 2007), and 0.6 0.6 near real-time acoustic monitoring for use in alerting 0.5 0.5 observing mariners to the presence and location of right whales is of

0.4 0.4 vessel−whale becoming a real possibility (Clark et al. 2007). Regard- a 0.3 0.3 less of the method used, for the transmission of whale of locations to mariners to be successful, mariners must 0.2 0.2

be willing and able to safely manoeuvre to avoid Probability 0.1 0.1 potential collisions. 0 0

There is little compelling evidence to show that right 0 0.2 0.4 0.6 0.8 1 Probability whales avoid approaching vessels (Vanderlaan & Tag- Probability of observing a vessel gart 2007, see also Panigada et al. 2006) and whales 30 1 b may be habituated to vessel noise and ignore it 0.9

25 (Nowacek et al. 2004). Technological methodologies 0.8 for alerting whales to the presence of vessels include 0.7 active acoustic devices (scare tactics). Such devices 20 collision have been successful in reducing incidental entangle- 0.6 ments of harbour porpoises (e.g. Kraus et al. 1997, 15 0.5 lethal Trippel et al. 1999, Culik et al. 2001) but we know of no a

0.4

similar results for large baleen whales. When Todd et of

10 0.3 al. (1992) used such devices to alert humpback Vessel speed (knots)

Megaptera novaeangliae and minke Balaenoptera 0.2 Risk 5 acutorostrata whales to the presence of fishing gear, 0.1 the animals approached closer to gear with active 0 0 devices than to gear with inactive devices. Nowacek et 0 0.2 0.4 0.6 0.8 1 al. (2004) demonstrated that right whales were not Probability of a vessel−whale encounter deterred by disharmonic alarm sounds spanning the presumed hearing range of the whales, and the alarm Fig. 1. Nomographs illustrating (a) the generalised 0,1 probability of observing a vessel and a whale at the same time and resulted in the whales swimming strongly to the sur- location, i.e. encounter, and (b) the risk of a lethal collision as face, where they were exposed to an increased likeli- a function of speed (from Vanderlaan & Taggart 2007) given hood of a vessel strike. a vessel–whale encounter

003

Vanderlaan et al.: Reducing vessel–whale lethal encounter risk 285

through October (Gaskin 1987, Stone et al. 1988,

Brown et al. 1995); the Grand Manan Basin in the Bay Bay of Fundy 45.0° USA of Fundy (hereafter Fundy) and Roseway Basin on the N (Maine) Grand Manan Basin southwest Scotian Shelf (hereafter Roseway). Each habitat encompasses a Canadian Right Whale Conser- vation Area (Fig. 2) that serves only to warn vessel crews transiting the regions that right whales are likely to be present (Brown et al. 1995). There are no vessel Canada (Nova Scotia) regulations (e.g. routing, speed restrictions) associated 44.0 with these conservation areas, and actions taken by a vessel crew to minimise a whale-strike, as recommended on nautical charts and notices to mariners Gulf of Maine (slow down and/or avoid region), would be strictly voluntary (Brown et al. 1995). Scotian Shelf

In the present study we quantitatively address the 43.0 Roseway problem of vessels striking whales by using vessel traf- Basin fic and right whale survey and sighting data to determine the relative probability of vessel and right whale encounters at ~3 nautical mile (n mile)1 resolution Nautical miles (~5 km) in the Fundy and Roseway regions. We do so, 0 Kilometres 50 200 in part, because the Fundy and Roseway regions are 00 0 1510 100 Georges Bank places where aggregations of right whales intersect 42.0 typical coastal vessel-traffic patterns: self-determined 68.0 67.0 66.0 65.0°W lanes in Roseway, and a formal traffic separation Fig. 2. Bathymetric (m) chart illustrating the Gulf of Maine, scheme in Fundy. Together with the encounter proba- the Canada Exclusive Economic Zone limit (thick black line), bility and vessel speed estimates, we employ the Van- and the Grand Manan (Bay of Fundy) and Roseway (western Scotian Shelf) basin study domains (red dashed line) that derlaan & Taggart (2007) lethality model to estimate encompass the Canadian Right Whale Conservation Areas the relative risk of a lethal encounter between a vessel (black dashed line) and a right whale in Fundy and Roseway. We then use the results to illustrate how these analyses have (IMO 2003), and will (IMO 2007a,b), be employed as conser- using a Cessna® 337 Skymaster® flying at a nominal vation practices to minimise the risk of lethal vessel 100 knots (185 km h–1) at 230 m altitude along E–W strikes to right whales. survey lines spaced at nominally 5 n mile (9.3 km) intervals (see also Scott & Gilbert 1982, CETAP 1982). Quality controlled right whale sightings per unit MATERIALS AND METHODS effort (SPUE, number of whales observed per 1000 km of the standardised survey track) data, collected during Whale data. An overview of the whale survey surveys described above for the period 1987 through methodologies and quality control are provided in 2000, were provided for each of the Fundy and Rose- Brown et al. (2007), though the salient features are way regions (no survey data for 1993, 1994, 1996, 1998 summarised here. Survey platforms were primarily in the latter) by the North Atlantic Right Whale Con- vessels and, secondarily, aircraft that followed sys- sortium (NARWC 2005; see also Kenney 2001). We tematic survey lines. Observers used standardised aggregated the annual data, as annual SPUE estimates methods and vessels travelled at a nominal 12 knots are too limited in time and space in any given year. (22 km h–1) along typically N–S survey lines spaced at The whale data were then resolved to represent cell- a nominal 4 n miles (7.4 km), and data used for analyt- specific SPUE estimates across the standard NARWC ical purposes were considered valid only when visibil- 20  20-cell (Fundy; area = 2520 n miles2, 8643 km2) ity was nominally 2 n miles (3.7 km) and in sea state and 25  20-cell (Roseway; area = 3300 n miles2, nominally < 4 (Beaufort wind force scale). All right 11 319 km2) grids with each cell defined by 3’ N lati- whales were counted and their locations geo-refer- tude and 3’ W longitude; i.e. 3 n miles (5.6 km) N–S enced. Aerial surveys were conducted as above and ~2.1 n miles (3.9 km, Fundy) and ~2.2 n miles (4.1 km, Roseway) E–W. This is the limiting resolution used for all analyses presented below. 1We use knots and nautical mile (n mile) units as they reflect Survey effort is not uniform across either of the the nautical convention, and we provide SI equivalents regional grids, and our calculations based on the SPUE

004

286 Endang Species Res 4: 283–297, 2008

data disregard biases that may be associated with the ing grid-cells because it did not occupy the cell(s) at geographic distribution of effort. One grid-cell in the the time the record was logged, or (2) a vessel had an Roseway grid had an anomalously high SPUE value incomplete (partially missing) track over an extended (2626), representing a near 4-fold increase over the period because there were no records (thus no track) next highest value. In comparison to the frequency dis- for the extended period. In either case, the vessel track tribution of all non-zero SPUE estimates (n = 66) over and associated data were interpolated (Euclidean) the grid, the value 2626 is ~16-fold greater than the across adjoining grid-cells. Unique-vessel tracking inter-quartile range, was considered an extreme out- algorithms were developed for quality control, and lier, and was replaced by the overall grid-cell mean where quality norms based on location, time and speed SPUE estimate (excluding the outlier) of 22.4. evolution comparisons were not met, a graphical pro- Vessel traffic: Bay of Fundy. Vessel-tracking radar cedure was employed to achieve the required inter- data for the period 9 June through 31 October 2000 in polation; otherwise the data were discarded. For ex- Fundy were extracted from the Kongsberg Norcontrol ample, some uniquely identified vessels appeared at IT® Vessel Traffic Management and Information Sys- more or less random locations in time for which no tem computerised log-files stored with the Marine rational interpretation or interpolation could be Communication and Traffic Services (MCTS) Saint inferred; such data were rare and were discarded. For John, New Brunswick, Canada. Each data record con- the most part, ill-behaved vessel data were attribut- tains various fields that include ‘rule-vessel’ (those able to ‘non-rule’ vessels. 20 m in overall length and 300 gross registered ton- All vessel data (identity, number, speed etc.) were nage, GRT; Canada Shipping Act 1985) identity, loca- then aggregated on a daily (24 h) basis across the 400 tion (latitude and longitude), date, time, and speed grid-cell domain and thus spatially resolved as above (knots) logged at 2.50  10–3 Hz. Only those vessel data for the SPUE data. This procedure provided the num- that fell within the domain of the NARWC survey grid ber of vessels transiting the cell and the weighted were used. There were some periods within some days mean (by unique vessel) vessel speed within the cell. when tracking data were not logged (min to h). There Some unique vessels entered a given cell more than were no data available for the entirety of 27 June. once within a 24 h period (return trips, backtracking) There were no data for the period 05 through 22 Au- and they were statistically treated as new observa- gust 2000 and thus the entire August 2000 dataset was tions. There were extremely rare occasions when a replaced by the entire August 2001 dataset. We as- rule vessel entered a grid-cell more than 20 times sume that the few missing data and the replacement within a day (typically research or coast-guard ves- data do not unduly compromise the analyses. This as- sels), and only the first 20 entries were used. For the sumption, for the August data, was validated through purposes of this study, the daily statistics were then an analysis of day-to-day rule-vessel movements aggregated through time, to estimate the total number (Transport Canada data) at major ports in the Bay of of vessels and mean speed per grid-cell for June Fundy over the period 2000 through 2002 inclusive. through October. The mean-speed estimates for 4 grid ‘Non-rule’ vessel-targets (e.g. fishing boats, pleasure cells (each with a land –water interface) were deter- craft, unidentified objects etc.) are also tracked period- mined to be statistical outliers (possibly related to ically and at the discretion of MCTS controllers, and land –water–vessel radar reflection effects) and these such radar data were ignored in our analyses due to speed estimates (not vessel numbers) were excluded the paucity of systematic data logging. MCTS con- from analyses. trollers assign rule-vessel names (identifiers) at their The Bay of Fundy Traffic Separation Scheme (TSS) discretion and thus there are many aliases for a given spatially separates inbound and outbound rule-vessels unique rule-vessel among the records. Accordingly, entering and departing the Bay (see Fig. 3). The extensive efforts to quality control and error-check International Maritime Organisation (IMO) adopted were made to ensure all aliases were identified and amendments to the existing TSS and they were imple- assigned to the appropriate ‘unique’ vessel. There mented on 01 July 2003 (IMO 2003). In the following were instances when the same apparently unique ves- we refer to the pre-amendment TSS as the ‘original’ sel was located in 2 different places at the same time; and the post-amendment TSS as the ‘amended’. in such cases the data were retained because each ves- Vessel traffic: Roseway Basin. Roseway vessel traffic sel was identified as a rule vessel (though of unknown data for 1989 to 2002 inclusive were available from the identity). Eastern Canada Vessel Traffic Services Zone Regula- Data extraction and assignment of vessel data to the tions (ECAREG; Canada Shipping Act 1985) database appropriate NARWC grid-cell was relatively straight- and from the International Comprehensive Ocean-At- forward. However, adjustments were required when mosphere Data Set (ICOADS, National Center for At- (1) a vessel appeared to ‘skip’ one or more non-adjoin- mospheric Research [NCAR], Boulder, Colorado).

005

Vanderlaan et al.: Reducing vessel–whale lethal encounter risk 287

ECAREG is a mandatory reporting system that requires where Vi is the aggregate vessel number occupying vessels (500 GRT or carrying pollutants or dangerous grid-cell i. cargo) to report (call in) 24 h prior to entering an Using Eqs. (1) & (2), the relative probability that a ECAREG zone (all Canadian waters south of 60° N and vessel and whale will occupy (encounter each other the St Lawrence River east of 66° W) and 2 h prior to de- within) a given grid-cell i is then calculated: parting a Canadian port in an ECAREG zone. Vessels Prel  Whalei  Prel  Vesseli P Encounter  report to ECAREG more frequently than required. rel i n (3) ICOADS is derived from the fleet of Voluntary Observ- Prel  Whalei  Prel  Vesseli  ing Ships and represents ~9% of the world commercial i 1 fleet (Corbett & Koehler 2003, Corbett 2004). The where the Prel (Encounter)i estimates are normalised, ECAREG data (exact location at reporting of unique such that their sum across the grid is equal to 1. vessels) were quality controlled for errors, such as im- The Prel (Encounter)i estimates in Fundy before and possible dates, locations, speeds, and tonnages, and after the TSS amendment were compared by using errors were treated as missing values. ICOADS data Eq. (4) below and by ‘moving’ those vessels transiting (unique vessels resolved to 6’ grid-cells) are quality the original lanes into the amended lanes. This was controlled at NCAR, with errors treated as missing val- achieved by identifying any vessel transiting an origi- ues. The ECAREG and ICOADS records were ex- nal lane, determining its longitudinal location and tracted and assigned to a large 32  30 6’ grid (centred adjusting that location to place the vessel in the appro- on the centre of the NARWC 25  20-cell Roseway 3’ priate amended lane at the same relative distance from grid) and summed to provide the total number of ves- the western boundary of the lane. The longitudinal sels reporting (vessel count) in each 6’ grid-cell. The locations of those vessels that exited or entered the mean speed of vessels within each grid-cell was calcu- original TSS to and from ports in Maine and southern lated based on the ECAREG reported speeds and the New Brunswick were adjusted as above and their mid-point of each 5 knot (9.3 km h–1) speed-class inter- routes were interpolated to allow such vessels to enter val provided by ICOADS. The vessel data resolved at 6’ and exit the ‘turnout lanes’ (implemented under the were then smoothed using a spline function (Surfer® TSS amendment; see Fig. 4b) after which their routes ver.8; 2002 Golden Software) to resolve the data at 3’. were interpolated across the grid to their originating or Only those data encompassed by the NARWC 25  20- terminating locations. cell 3’ grid were used in the analyses. To compare relative encounter probabilities before For both ECAREG and ICOADS, the vessel data rep- and after the TSS amendment in Fundy and between resent only those vessels that reported within the the Fundy and Roseway regions, Eq. (3) is modified to: above-specified 3’-grid domain and not the total num- Prel  Whalei  Prel  Vesseli ber of vessels that transit the domain. However, the Prel Encounteri  m (4) extent of the data (1989 through 2002) for both sets of Prel  Whalei  Prel  Vesseli  data is sufficient to provide statistically reliable esti- i 1 mates of the spatial distribution of vessel traffic and where m = (noriginal + namended) or m = (nFundy + nRoseway), vessel speeds. respectively. Estimating relative probability of encounter. We Estimating relative risk. Using the model provided assume the 1987 to 2000 aggregate SPUE estimate in Vanderlaan & Taggart (2007) and the mean vessel (SPUEi) provides the best estimate of the relative prob- speed within a grid-cell, the probability of a lethal ability, at 3’ resolution, that a whale occupies a grid- injury (given encounter) is estimated for both regions. cell i relative to other cells in a domain of n cells (sim- We then quantify the relative risk (RRi), at 3’ resolution, plification of the 2-dimensional nx,y grid) and is of vessels to right whales based on the event: the rela- calculated as: tive probability of a vessel encountering a whale (Eq. 3) and the consequence: the probability that the SPUEi Prel Whalei  n (1) encounter is lethal: SPUEi 1 i 1 P(Lethal | Encounter)i  (5) 1  exp(4.89  0.41xi ) Similarly, the relative probability that a vessel occu- – pies a grid-cell i relative to other cells in a domain of n where xi is the mean vessel speed (knots) in grid-cell i. cells is calculated as: Thus, the relative risk becomes:

RRi  Prel(Encounter)i  P(Lethal | Encounter)i (6) Vi Prel Vesseli  (2) n Spatial distribution data (e.g. Prel (Whale), Prel (Ves- Vi i 1 sel), Prel (Encounter), RR etc.) are presented as objec-

006

288 Endang Species Res 4: 283–297, 2008

tive contours (Surfer®, ver.8; 2002 Golden Software) the study period. The elevated Prel(Vessel) in both that rely primarily on kriging and natural neighbour the inbound and outbound lanes of the original TSS, algorithms. All chart figures are equidistance cylindri- where it bends from NE to NNE just south of the cal projection. All statistical uncertainties are pre- Conservation Area (the 3A and 3B MCTS call-in sented as ±1 standard error. locations), reflects a traffic node where vessels enter the lanes from the NW (coastal Maine and southern New Brunswick) and from the ENE (north shore of RESULTS Nova Scotia).

Right whales in the Bay of Fundy Relative probability of a vessel encountering a Based on the NARWC grid domain, there is an over- right whale in the Bay of Fundy all 67% chance of observing a right whale within the Grand Manan Basin region that is encompassed by the It is clear, and not surprising, that the highest rela- Right Whale Conservation Area — an estimate that tive probability of a vessel encountering a right corresponds to a mean relative probability of 0.023 ± whale, on the basis of aggregated vessel number, is 0.0029 that is, on average, ~36-fold greater than else- located in the NE sector of the Grand Manan Basin where in the domain (Fig. 3a). There is a second and and in the Conservation Area where it is intersected smaller area of slightly elevated relative probability by the outbound lane of the original TSS (Fig. 3c). north of the Conservation Area and it is bounded to the There are also elevated relative encounter probabili- north by the 100 m isobath. There is another and larger ties to the NNE and S of the Conservation Area — elevated relative probability immediately south of the reflecting the whale distributions (Fig. 3a), given the Conservation Area, possibly reflecting ingress and relatively uniform distribution of vessels in the TSS- egress of the whales to and from the Grand Manan lanes (Fig. 3b). Basin region.

Vessel speed in the Bay of Fundy Vessels in the Bay of Fundy The mean and median speeds of rule vessels in A total of 768 unique rule-vessels were identified the Fundy domain during the study period were within the MCTS data. A further 759 vessel identities ~12 knots (~22 km h–1) and ~11 knots (~20 km h–1) were variants (aliases) of identifiable rule vessels that respectively (Table 1). Vessel speeds were greatest were corrected to their unique identities. Of these SW of the seaward entrance to the original TSS unique vessels, a total of 42 were included in the and remain relatively high (in excess of 14 knots, analyses despite being identified as being in 2 differ- 26 km h–1) where the TSS intersects the Right Whale ent places as the same time. Thus, the analyses are Conservation Area (Fig. 3d). There is no clear evi- based on a total of 1485 rule vessels, many of which dence that vessels reduce speed when navigating made several trips through the study domain during through the Conservation Area, as recommended the June through October period. There were at least on nautical charts. 2622 unidentified, non-rule, vessel-targets that were excluded from our analyses. –1 –1 Table 1. Summary vessel speed (in knots, with speed in The average number of rule vessels grid-cell d km h–1 in parentheses) statistics across the 3’ grid-cell do- in the Fundy domain was 0.37 ± 0.028. The grid-cell- mains in the Bay of Fundy (based on radar data) and Roseway averaged number of vessels within the inbound and Basin (based on Eastern Canada Vessel Traffic Services Zone outbound lanes of the original TSS was 1.1 ± 0.076 Regulations, ECAREG, and International Comprehensive vessels cell–1 d–1. Approximately 71% of all rule ves- Ocean-Atmosphere Data Set, ICOADS data). n = no. of grid-cells occupied by vessels sels were transiting the TSS-lanes resulting in a –4 mean Prel(Vessel) of 0.0072 ± 5.2  10 within the TSS-lanes (Fig. 3b). The remaining 29% were navi- Statistic Bay of Fundy Roseway Basin gating elsewhere within the domain. The mean n 329 500 –4 Prel(Vessel) outside the TSS-lanes was 9.7  10 ± Mean 11.6 (21.5) 11.2 (20.7) 7.2  10–5; i.e. rule vessels were on average 7.4-fold Standard deviation 3.12 (5.78) 1.5 (2.8) more likely to be within the TSS-lanes than not. Minimum 0.614 (1.14) 7.0 (13.0) Median 11.3 (20.9) 10.9 (20.2) Approximately 18% (454 n mile2, 1556 km2) of the Maximum 24.5 (45.4) 15.3 (28.3) Fundy domain was unoccupied by a rule vessel over

007

Vanderlaan et al.: Reducing vessel–whale lethal encounter risk 289

45.2° a b N 0.015 0.045 0.014 45.0 0.040 0.013 0.012

0.035 0.011 44.8 0.010 0.030 0.0090 44.6 0.025 0.0080 0.0070 0.020 0.0060 44.4 0.015 0.0050 0.0040 0.010 0.0030 44.2 Nautical miles Nautical miles

0 14 0 14 0.0020 Kilometres 0.0050 Kilometres 0.0010 0 20 0 20 44.0 0.0010 0

45.2° c d N 0.075 24.0 0.070 45.0 0.065 17.5

0.060 15.5

44.8 0.055 14.0 0.050 0.045 13.0 44.6 0.040 12.0 0.035 11.0 44.4 0.030 0.025 10.0

0.020 8.80 44.2 Nautical miles Nautical miles 0.015 0 14 0 14 Kilometres Kilometres 6.80 0.010 0 20 0 20 44.0 0.0050 0 67.2 67.0 66.8 66.6 66.4 66.2 66.0 65.8°W 45.2° N e 0.055

45.0 0.050

0.045

44.8 0.040

0.035

44.6 0.030 Fig. 3. Bathymetric (100 m resolution) charts of the Bay of 0.025 Fundy illustrating the study domain (red dashed line), Cana- 44.4 0.020 dian Right Whale Conservation Area (black dashed line), and original traffic separation scheme (solid black line) (inbound 0.015 lane to E and outbound lane to W) and showing the relative probability of (a) observing a right whale, (b) observing a ves- 44.2 Nautical miles 0.010 sel, (c) a vessel encountering a right whale, and (d) average

0 14 vessel speed (knots) and (e) relative risk of a lethal collision Kilometres 0.0050 between a vessel and a right whale. Note panel (d) is colour- 0 20 44.0 0.0010 scaled to match that of lethal collision as a function of vessel 67.2 67.0 66.8 66.6 66.4 66.2 66.0 65.8°W speed shown in Fig. 1b

008

290 Endang Species Res 4: 283–297, 2008

From the above, it becomes apparent that when the 45.2° concentration of vessels in the original TSS-lanes N a 0.035 (Fig. 3b) is coupled with the vessel-speed estimates 0.033 (Fig. 3d) the greatest risk to the right whales is 45.0 expected to be in that part of the Conservation Area 0.030 associated with elevated SPUE estimates (Fig. 3a). 0.028

44.8 0.025

0.023 Probability of a lethal injury and relative risk in the Bay of Fundy 0.020 44.6 0.018 When the probability-of-lethal-injury model (Vander- 0.015 laan & Taggart 2007) is applied to the spatial estimates of vessel speeds, we estimate an average P (Lethal|En- 44.4 0.013 counter) of 0.64 within the original TSS-lanes of the Bay 0.010 of Fundy; i.e. if a whale was struck within the original 0.0075 TSS-lanes there is a 64% chance, on average, that it 44.2 Nautical miles 0.0050 would be killed. This probability increases as vessels in- 0 14 Kilometres crease their speed near the seaward end of the TSS 0.0025 0 20 (Fig. 3d). If a whale is struck in the area between the coast 44.0 0.0010 of Maine and Grand Manan Island, New Brunswick, the 45.2° probability of a lethal injury is also relatively high with N b P (Lethal|Encounter) ranging from 0.72 to 0.84. 0.035 The average relative risk to a right whale from vessels 0.033 45.0 travelling through the original TSS-lanes is 0.0057 0.030 (Fig. 3e). This estimate represents a 27-fold increase in 0.028 the average probability of a whale being struck and killed in the TSS-lanes over anywhere else in the grid 44.8 0.025 domain. The average relative risk to right whales found 0.023 only in the portion of the Conservation Area intersected 0.020 by the outbound lane of the original TSS outbound lane 44.6 0.018 is 270-fold greater than elsewhere in the domain.

0.015 44.4 0.013

TSS amendment in the Bay of Fundy 0.010

0.0075 The overall relative probability of a vessel encounter- 44.2 Nautical miles 0.0050 ing a right whale in the Fundy grid domain with the orig- 0 14 Kilometres inal TSS decreases by 44% with the amended TSS. For 0.0025 0 20 those 5 grid-cells within the outbound lane of the original 44.0 0.0010 TSS that intersects the Right Whale Conservation Area, 67.2 67.0 66.8 66.6 66.4 66.2 66.0 65.8°W there is an average reduction of 90% ± 4.2 in the relative Fig. 4. Bathymetric (100 m resolution) chart of the Bay of encounter probability associated with the amendment. Fundy illustrating the study domain (red dashed line), Cana- The mean relative probability of encounter over the en- dian Right Whale Conservation Area (black dashed line), the tire TSS-lanes decreased by ~40% (from 0.0080 ± 0.0019 original (a) and amended (b) traffic separation schemes (solid to 0.0049 ± 0.00098) with the TSS amendment. The stan- black line) and their associated standardised (comparable scale) relative risk of a lethal collision between a vessel and dardised overall relative risk to right whales of a lethal a right whale vessel-strike in the Fundy domain decreased by 62% with the TSS amendment (Fig. 4). The average relative risk to the whales over the entire TSS-lanes also de- Right whales in Roseway Basin creased by ~40% (from 0.0057 ± 0.0015 to 0.0034 ± 7.3  10–4) with the amendment. For those 5 grid-cells within Based on the NARWC grid domain, there is an the outbound lane of the original TSS that intersected the overall 86% chance of observing a right whale within Right Whale Conservation Area, there was an average the Roseway Basin Right Whale Conservation Area — reduction of 90% ± 4.7 in risk with the amendment. an estimate that corresponds to a mean relative prob-

009

Vanderlaan et al.: Reducing vessel–whale lethal encounter risk 291

44.0° a b N 0.060 0.0056 43.8 0.055 0.0052

0.050 43.6 0.0048 0.045 0.0044 43.4 0.040 0.0040 0.035 0.0036 43.2 0.030 0.0032 43.0 0.025 0.0028 0.020 0.0024 42.8 0.015 0.0020 Nautical miles 0.010 Nautical miles 0.0016 42.6 0 20 0 20 Kilometres 0.0050 Kilometres 0.0012 42.4 0 30 0.00010 0 30 0.00080

44.0° N c d 0.060 24.0 43.8 0.055 17.5 0.050 43.6 15.5 0.045 43.4 0.040 14.0 0.035 13.0 43.2 0.030 12.0 43.0 0.025 11.0 0.020 10.0 42.8 0.015 8.80 Nautical miles 0.010 Nautical miles 42.6 0 20 0 20 6.80 Kilometres 0.0050 Kilometres 42.4 0 30 0.00010 0 30 0 66.6 66.4 66.2 66.0 65.8 65.6 65.4 65.2 65.0 64.8 64.6 64.4°W 44.0° N e 0.036 43.8 0.033

43.6 0.030 0.027

43.4 0.024 0.021 43.2 0.018 Fig. 5. Bathymetric (100 m resolution) chart of the Roseway Basin region illustrating the study domain (red dashed 43.0 0.015 line), Canadian Right Whale Conservation Area (black 0.012 dashed line), and showing the relative probability of (a) observing a right whale, (b) observing a vessel, (c) a vessel 42.8 0.0090 encountering a right whale, and (d) average vessel speed Nautical miles 0.0060 (knots) and (e) relative risk of a lethal collision between a 42.6 0 20 vessel and a right whale. The recommendatory area to be Kilometres 0.0030 avoided is outlined (solid black line). Note panel (d) is

42.4 0 30 0.00030 colour-scaled to match that of lethal collision as a function 66.6 66.4 66.2 66.0 65.8 65.6 65.4 65.2 65.0 64.8 64.6 64.4°W of vessel speed shown in Fig. 1b 010

292 Endang Species Res 4: 283–297, 2008

ability of 0.010 ± 0.0017 that is on average ~26-fold Vessel speed in Roseway Basin greater than elsewhere in the domain (Fig. 5a). The highest relative probability of observing a right whale The mean and median speeds of rule vessels across in the Basin occurs in the south-central region of the the Roseway domain during the study period are both –1 Conservation Area (Prel(Whale) = 0.0707). There are ~11 knots (21 km h ) with a maximum grid-cell mean also small regions of elevated probability to the E and speed of 15 knots (28 km h–1; Table 1). The spatial dis- to the W. tribution of the average vessel speeds across the grid is similar to the spatial distribution of vessels (Fig. 5d). However, the diagonal traffic pattern that intersects Vessels in Roseway Basin the Right Whale Conservation Area is associated with vessels navigating at greater speeds (13 to 15 knots: 20 There were 5374 ECAREG call-in locations in the to 28 km h–1) than vessels to the north where they are Roseway domain over the 1989–2002 period, repre- most concentrated and are generally navigating at senting 501 unique vessels based on the Lloyd’s Regis- speeds of 10 to 11 knots (19 to 20 km h–1). There is no ter number and an additional 605 records where the clear evidence that vessels reduce speed when navi- Lloyd’s number was either not reported or not gating through the Conservation Area, as recom- recorded. Over the same period, there were 3877 mended on nautical charts. ICOADS vessel-call-in locations in the domain, representing 466 unique vessels based on call sign and an additional 10 records where the call sign was not Probability of a lethal injury and relative risk in reported or not recorded. The unidentified vessels in Roseway Basin either data set were included in the analyses and when the combined data were re-resolved to 3’ there were a The probability of a lethal vessel-strike (arising from total of 9036 call-in locations in the domain over the an encounter) to a right whale in the Roseway domain study period (~2% loss in re-resolving). ranges from 0.12 to 0.80, with an overall mean proba- The mean relative probability of observing a vessel bility of 0.43 ± 0.0064. The average probability of a –5 in the domain, Prel(Vessel), is 0.0020 ± 3.8  10 lethal injury within the Right Whale Conservation (Fig. 5b). The most concentrated vessel traffic in the Area is marginally higher at 0.49 ± 0.016. The mean region (Prel(Vessel)  0.0050) is located north of the relative risk to right whales from vessels over the –4 Conservation Area. Elsewhere in the domain Prel(Ves- entire Roseway domain is 0.0012 ± 2.4  10 (Fig. 5e). sel) ranged from 0.0014 to 0.0023 (1st to 3rd quartile). Due to the aggregation of right whales within the Con- An emergent diagonal traffic pattern associated with servation Area and the intersecting traffic navigating an elevated Prel(Vessel)  0.0025 intersects the Right at relatively high speed, the mean relative risk is ~68 Whale Conservation Area in a NESW direction and times higher at 0.0060 ± 0.0011. indicates that few vessels navigating NESW avoid the Conservation Area, as recommended on nautical charts. DISCUSSION

The greatest probability of observing a North Relative probability of a vessel encountering a Atlantic right whale occurs within the Canadian Right right whale in Roseway Basin Whale Conservation Areas in each of the Grand Manan and Roseway basins. It is within these areas In comparison to the Conservation Area in Fundy, that vessels pose the greatest risk to right whales. In the relatively uniform distribution of vessel traffic in each conservation area the whales are exposed to ves- the Roseway Basin Conservation Area (Fig. 5b) sels navigating at speeds near to, or in excess of results in the relative encounter probabilities be- 13 knots (24 km h–1), corresponding to a P (Lethal| tween a vessel and right whale (Fig. 5c) generally Encounter) of at least 0.6. For the Bay of Fundy we reflecting the relative probability of observing a have shown that the 2003 TSS amendment decreased whale (Fig. 5a); i.e. the highest relative probability of the overall probability of a vessel encountering a right a vessel encountering a right whale is where the whale by 44% and the overall risk by 62%, and these highest probability of a observing a whale is located. estimates include reductions achieved through the use The mean relative probability of a vessel encounter- of the ‘turn-out’ lanes just north of the Conservation ing a right whale is 36 times higher within the Con- Area (Fig. 6a). In the region where the outbound lane servation Area than it is for the study domain outside of the original TSS intersected the highest expected the Conservation Area. concentrations of right whales, the TSS amendment

011

Vanderlaan et al.: Reducing vessel–whale lethal encounter risk 293

45.2° expected if vessel operators comply with the avoidance a N 0.0050 recommendation. 0.0025 We can use the above risk-reduction estimate to 45.0 0 coarsely estimate the actual number of documented lethal vessel–whale collisions that might accrue due to –0.0025 modified vessel-traffic patterns. Three right whale –0.0050 44.8 deaths were attributed to vessel strikes in the Bay of –0.0075 Fundy during the period 1990 though July 2003 –0.010 (Knowlton & Brown 2007); an average annual mortality –0.013 44.6 rate of 0.24 (1 every 4 yr). A 62% reduction in this rate –0.015 leads to the expectation of 0.081 lethal vessel strikes –0.018 per year (1 every ~12 yr). Although we cannot calcu- 44.4 –0.020 late a reasonable estimate of risk reduction in the –0.023 Roseway region because it is a function of unknown (at –0.025 least at this time) vessel-operator compliance, we can

44.2 Nautical miles –0.028 offer similar calculations assuming a 25, 50, and 75%

0 14 –0.030 reduction in risk entirely attributable to modified ves- Kilometres –0.033 sel navigation. One right whale death was attributed to 0 20 44.0 –0.035 vessel strike in the Roseway Basin between 1990 and 2005 inclusive; corresponding to an average annual 45.2° documented mortality rate of 0.063 (1 every 16 yr). b N 0.0050 Assuming a 25, 50, and 75% reduction in risk through 0.0025 vessel compliance the estimated number of docu- 45.0 0 mented lethal vessel strikes would be approximately 1 –0.0025 every 21, 32, and 64 yr, respectively. We emphasise –0.0050 that the above estimates are exceedingly conservative, 44.8 as they rely only on observed (documented) lethal ves- –0.0075 sel strikes; i.e. it is estimated that only 17% of all right –0.010 mortalities are observed (Kraus et al. 2005). –0.013 44.6 Technological methodologies aside (see ‘Introduc- –0.015 tion’), the most pragmatic means of reducing vessel –0.018 strikes to whales are to (1) reduce the probability of a 44.4 –0.020 vessel encountering a whale through modified vessel –0.023 routing (seasonal or otherwise), (2) reduce the lethality –0.025 of vessel strikes, should a collision occur, through ves-

44.2 Nautical miles –0.028 sel-speed restrictions, and (3) reduce overall risk

0 14 –0.030 through modified routing coupled with speed restric- Kilometres –0.033 tions. In the Bay of Fundy, we have the opportunity to 0 20 44.0 –0.035 determine the relative merits of each of the above 67.2 67.0 66.8 66.6 66.4 66.2 66.0 65.8°W means of reducing risk by comparing standardised risk Fig. 6. Bathymetric (100 m resolution) charts of the Bay of estimates based on Eqs. (4) & (6) above, and by con- Fundy illustrating the study domain (red dashed line), Cana- straining the average vessel speed in a grid-cell to a dian Right Whale Conservation Area (black dashed line) and maximum of 10 knots (18.5 km h–1). the standardised (comparable scale) residual risk of lethal col- Relative to the original TSS, the amended TSS lision between a vessel and a right whale associated with (a) the original and amended (solid black line) traffic separation provides an overall risk reduction of 62% (Fig. 6a, scheme and (b) the original (solid black line) traffic separation Table 2), whereas a maximum 10 knot (18.5 km h–1) scheme with a 10 knot speed restriction imposed over the en- speed restriction over the entire study domain results tire study domain. Negative residuals indicate reduced risk in a risk reduction of 52% (Fig. 6b, Table 2). The

amended TSS provides a 10% greater reduction in the reduced both the vessel–whale encounter probability overall relative risk than would speed restrictions and the vessel risk to the whales by 90%. Given the implemented throughout the domain, especially within pending implementation of the IMO-adopted sea- the Conservation Area, and it has a lower impact on sonal-recommendatory area to be avoided (ATBA) in transit times for vessels navigating throughout the the Roseway Basin region, similar reductions in ves- Fundy domain. The outbound lane of the amended sel–whale encounter probabilities and risk are to be TSS is ~0.8 n miles (1.5 km) longer than the original

012

294 Endang Species Res 4: 283–297, 2008

Table 2. Per cent reduction in overall risk of lethal vessel collisions to North tion in risk. For example, other large Atlantic right whales estimated for various modified vessel-traffic options in whales including humpback, fin Bal- the Bay of Fundy relative to the original (prior to 1 July 2003) traffic separation scheme (TSS) aenoptera physalus, sei B. borealis and minke frequent the Bay of Fundy and generally reach maximum abun- Modified vessel-traffic options % reduction in overall risk Entire Conservation dance during the July to September domain Area period (Gaskin 1983). However, traffic-routing amendments designed to Amended TSS 62 – protect the right whale should not Speed 10 knots (18.5 km h–1) 52 7.5 have negative consequences on the Amended TSS and speed 10 knots (18.5 km h–1) 75 69 other whale species, especially if they are not sympatric with the right whales. Thus, the probability of lane, and based on a 12 knot (22.2 km h–1) average ves- observing one of these other species in a given area sel speed (Table 1), the increased length of the lane may be very different from that for right whales. We amounts to a 4 min increase in lane-transit time. This therefore used the methods described above along 1.4% increase in transit time achieves a 62% reduction with the SPUE data for fin, sei, humpback, and minke in risk. Furthermore, it takes less time (~5 h) for a ves- whales to estimate the risk to these species posed by sel to transit the amended outbound lane at the 12 knot vessels over the entire study domain in terms of the (22.2 km h–1) average speed than it would for a vessel original and amended TSS. For each species the TSS to transit the original lane under a 10 knot (18.5 km h–1) amendment achieved a decrease in overall risk; 28% speed restriction (~6 h). for minke, 27% for fin, 17% for sei, and 9% for hump- If the speed restrictions were applied only to the part back whales. Thus, the TSS amendment in the Bay of of the original TSS-lanes that intersects the Conserva- Fundy decreased the relative risk from vessels to virtu- tion Area, a small reduction (7.5%) in relative risk is ally all large whales that frequent the Bay. achieved although there is at least a 16 min (20%) The vessel-traffic patterns in the Roseway Basin increase in transit time (10 vs. 12 knots, 18.5 vs. region are quite different from those in the Bay of 22.2 km h–1). Thus, in the Bay of Fundy, a modified Fundy, primarily due to the TSS in the latter. There are routing accrues much greater risk reduction than does emergent traffic-lanes in the Roseway Basin domain a 10 knot (18.5 km h–1) speed limitation. The last option with the most concentrated traffic transiting north of of a combined TSS amendment and a 10 knot (18.5 km the Conservation Area and less concentrated traffic h–1) speed restriction over the entire study domain pro- transiting in a NESW direction through the Conser- vides a relative risk reduction of 75% compared to the vation Area and navigating at higher average speeds. 62% achieved via the 2003 TSS amendment alone, Due to differences in the nature of the vessel data, the though at considerable time/speed costs and a limited relative probabilities of observing vessels cannot be reduction of risk from those vessels navigating well directly compared between the Bay of Fundy and outside the Conservation Area. If the 10 knot (18.5 km Roseway Basin domains. However, by standardising h–1) speed restriction was applied only to the amended the relative encounter and risk probabilities (Eqs. 4 & TSS routing within the Conservation Area, the relative 6) we can compare the 2 regions. risk reduction would be 69%, a marginal increase over Prior to the TSS amendment, the relative encounter the 62% above. In summary, the greatest reduction in probability over the Fundy study domain was greater risk of lethal vessel-strikes to right whales in the Bay of than over the Roseway domain by a factor of 3.0. After Fundy would be achieved by reducing the vessel the TSS amendment this factor decreased to 1.7. Prior speeds to 10 knots (18.5 km h–1) or less while navigat- to the TSS amendment, the relative risk associated ing through the region in conjunction with the with the right whale Conservation Area in Fundy was amended TSS. However, the least time-cost to ship- 7.0-fold greater, on average, than the relative risk in ping interests is associated with the small change in the right whale Conservation Area on Roseway. After navigation that results from the amended TSS that the TSS amendment this factor decreased, on average, achieves reduced vessel–whale encounter probabili- to 1.7. ties. Although there is no TSS in the Roseway Basin The TSS amendment in the Bay of Fundy was imple- region, a seasonal (June through December) recom- mented to reduce risk to right whales in the Grand mendatory area to be avoided (ATBA, Fig. 5e) has Manan Basin where they are most concentrated during been proposed (IMO 2007a), approved (IMO 2007b) summer and autumn. Consequently, other large and will be implemented on 1 May 2008. The designa- whales inhabiting the same region also accrue a reduction of the ATBA has been achieved, in part, because

013

Vanderlaan et al.: Reducing vessel–whale lethal encounter risk 295

of the analyses provided above. The IMO-sanctioned Our analyses rely on the temporally aggregated vessel-avoidance scheme can reduce the probability of SPUE and vessel data that do not incorporate the daily, vessel encounters with right whales and can reduce seasonal and interannual viability in whale distribu- the risk of lethal vessel-collisions without the imposi- tions and vessel-traffic patterns, and we are thus tion of speed restrictions. The magnitude of the risk assuming that the spatial probabilities associated with reduction will be a function of vessel operator compli- both the whales and the vessels in either habitat area ance. As compliance with the Roseway ATBA is recom- remain stable. If measures are to be implemented to mendatory (voluntary), we cannot, at this time, easily protect whales (or other marine animals), designing estimate what the actual reduction might be. We could modifications to traffic patterns could be seasonally make calculations based on various anticipated area based as is the case for the ATBA in Roseway Basin; avoidance and compliance scenarios, but suffice to annually for the June through December period. How- say that compliance by a large proportion of the tran- ever, if seasonal and/or interannual temporal variabil- siting fleet will result in a substantial reduction in risk ity in animal distributions is large, then risk reduction given that the ATBA completely surrounds the region may require year-round implementation of traffic of highest probability of encountering a right whale restrictions. (cf. Fig. 5c). We can anticipate a high degree of com- The methods presented here are used to quantify pliance based on the assumption that most vessels relative encounter probabilities and relative risk to transiting the Conservation Area in a NESW direc- North Atlantic right whales from ocean-going vessels. tion at a 14 knot (26 km h–1) average speed (see They also provide a means of estimating reductions in Fig. 5b,d) will comply, given the small adjustment in relative encounter probabilities associated with the routing and travel time if avoidance-routing is planned various options available to reduce risk. Speed restric- ahead of time. For example, if vessels normally navi- tions and areas to be avoided are being proposed in gating either SW from Halifax or NE from the New the United States of America (NOAA 2006, Merrick & York City TSS (thus in the diagonal traffic pattern illus- Cole 2007) and elsewhere, and such analyses can be trated in Fig. 5b,d) were to adjust their heading by 7° used to directly assess the utility of re-routing, speed (SW or NE as required) at a point some 93 n miles restrictions or both. (172 km) either side of the ATBA boundary, they would transit just outside of the SW corner of the ATBA and the route would be 2 n miles (3.7 km) longer than their Acknowledgements. We thank the New England Aquarium, normal 226 n miles (419 km) route. At an average the North Atlantic Right Whale Consortium and the many –1 staff and volunteers who collected survey data, F. Webster speed of 14 knots (26 km h ) the increase in transit and D. Edmonds (MCTS Saint John, New Brunswick) and G. time would amount to 8.6 min over the now 228 n miles Herbert (Oceans and Coastal Management Division, Fish- (422 km) route. If a 10 knot (18.5 km h–1) speed restric- eries & Oceans Canada) for data access. We are grateful for tion was imposed for vessels inside and transiting the funding in support of the various initiatives addressed in the paper; funds were provided by S. Haney and the Canadian Conservation Area along the diagonal traffic pattern Whale Institute, the Environment Canada Habitat Steward- (as opposed to an ATBA), then the normal transit time ship Programme, the World Wildlife Fund Canada and Envi- of 2.3 h at an average 14 knots (26 km h–1) would ronment Canada Endangered Species Research Fund, Fish- increase by 40% to 3.2 h. As in the Bay of Fundy, eries & Oceans Canada (St. Andrews Biological Station and the Oceans and Coastal Management Division) and the Nat- rerouting of vessels in this region comes with lower ural Sciences and Engineering Research Council of Canada. costs to vessel operators than would a 10 knot (18.5 km We thank J. Beaudin Ring, D. Gillis, N. Helcl, J. Mullarney, A. h–1) speed restriction, while at the same time substan- Neuheimer, R. Pelot, B. Smith, C. C. Smith, and 2 anonymous tially reducing risk to the whales. A 10 knot (18.5 km referees for advice, critiques and assistance. h–1) speed restriction within the Conservation Area could serve as an incentive for vessel operators to comply with the ATBA. LITERATURE CITED To reduce the risk to whales, the goals of ATBAs, amended TSSs and/or speed restrictions should be to Baumgartner MF, Mayo CA, Kenney RD (2007) Enormous carnivores, microscopic food, and a restaurant that’s hard achieve the greatest reduction in risk balanced by to find. In: Kraus SD, Rolland RM (eds) The urban whale: some minimal disruption to vessel operations while North Atlantic right whales at the crossroads. Harvard maintaining safe navigation. Although Elvin & Taggart University Press, Cambridge, MA, p 138–171 (2008) argue that mitigating encounters between large Brown MW, Allen JM, Kraus SD (1995) The designation of vessels and right whales may be difficult, we have seasonal right whale conservation areas in the waters of Atlantic Canada. In: Shackell NL, Willison JHM (eds) shown it is feasible where the data are available. Sta- Marine protected areas and sustainable fisheries. Science tistical and probabilistic methods can be used to and Management of Marine Protected Areas Association, achieve the above goals, as we have shown. Wolfville, NS, p 90–98

014

296 Endang Species Res 4: 283–297, 2008

Brown MW, Kraus SC, Slay CK, Garrison LP (2007) Survey for Knowlton AR, Brown MW (2007) Running the gauntlet: right discovery, science, and management. In: Kraus SD, Rol- whales and vessel strikes. In: Kraus SD, Rolland RM (eds) land RM (eds) The urban whale: North Atlantic right The urban whale: North Atlantic right whales at the whales at the crossroads. Harvard University Press, Cam- crossroads. Harvard University Press, Cambridge, MA, bridge, MA, p 105–137 p 409–435 Campbell-Malone R, Barco SG, Daoust PY, Knowlton AR, Knowlton AR, Kraus SD (2001) Mortality and serious injury of McLellan WA, Rotstein DS, Moore MJ (2008) Gross and northern right whales (Eubalaena glacialis) in the western histologic evidence of sharp and blunt trauma in North North Atlantic Ocean. J Cetacean Res Manag 2(Spec Atlantic right whales (Eubalaena glacialis) killed by ves- Issue):193–208 sels. J Zoo Wildl Med 39:37–55 Knowlton AR, Kraus SD, Kenney RD (1994) Reproduction in Canada Shipping Act (1985) R.S., 1985, c.S-9; 2001, c.26 Reg- North Atlantic right whales (Eubalaena glacialis). Can J ulatory Reform. Act repealed 01 July 2007 in accord with Zool 72:1297–1305 SI/2007–65, 27 June, 2007. Canada Gazette Part II, Vol- > Kraus SD (1990) Rates and potential causes of mortality in ume 141, No. 10, 16 May 2007, Ottawa North Atlantic right whales (Eubalaena glacialis). Mar > Caswell H, Fujiwara M, Brault S (1999) Declining survival Mamm Sci 6:278–291 probability threatens the North Atlantic right whale. Proc Kraus SD, Rolland RM (2007) Right whales in the urban Natl Acad Sci USA 96:3308–3313 ocean. In: Kraus SD, Rolland RM (eds) The urban whale: CETAP (Cetacean and Turtle Assessment Program) (1982) A North Atlantic right whales at the crossroads. Harvard characterization of marine mammals and turtles in the University Press, Cambridge, MA mid- and North Atlantic areas of the U.S. outer continental > Kraus SD, Read AJ, Solow A, Baldwin K, Spradlin T, Ander- shelf. Final Report of the Cetacean and Turtle Assessment son E, Williamson J (1997) Acoustic alarms reduce por- Program to the U.S. Department of the Interior, Washing- poise mortality. Nature 388:525 ton, DC. Contract AA5551-CT8–48 Kraus SD, Hamilton PK, Kenney RD, Knowlton AR, Slay CK Clark CW, Gillespie D, Nowacek DP, Parks SE (2007) Listen- (2001) Reproductive parameters of the North Atlantic right ing to their world: acoustics for monitoring and protecting whale. J Cetacean Res Manag 2(Spec Iss):231–236 right whales in an urbanized ocean. In: Kraus SD, Rolland > Kraus SD, Brown MW, Caswell H, Clark CW and others RM (eds) The urban whale: North Atlantic right whales at (2005) North Atlantic right whales in crisis. Science the crossroads. Harvard University Press, Cambridge, 309:561–562 MA, p 333–357 > Laurinolli MH, Hay AE, Desharnais F, Taggart CT (2003) Corbett JJ (2004) Marine transportation and energy use. In: Localization of North Atlantic right whale sounds in the Cleveland CJ (ed) Encyclopedia of energy, Vol 3. Elsevier Bay of Fundy using a sonobuoy array. Mar Mamm Sci Science, San Diego, CA, p 745–748 19:708–723 > Corbett JJ, Koehler HW (2003) Updated emissions from ocean Matthews JN, Brown S, Gillespie D, Johnson M and others shipping. J Geophys Res 108(D20):4650–4666 (2001) Vocalisation rates of the North Atlantic right whale > Culik BM, Koschinski S, Tregenza N, Ellis GM (2001) Reac- (Eubalaena glacialis). J Cetacean Res Manag 3:271–282 tions of harbor porpoises Phocoena phocoena and herring > Mellinger DK, Nieukirk SL, Matsumoto H, Heimlich SL and Clupea harengus to acoustic alarms. Mar Ecol Prog Ser others (2007) Seasonal occurrence of North Atlantic right 211:255–260 whale (Eubalaena glacialis) vocalizations at two sites on Elvin S, Taggart CT (2008) Right whales and vessels in Cana- the Scotian Shelf. Mar Mamm Sci 23:856–867 dian waters. Mar Pol 32:379–386 Merrick RL, Cole TVN (2007) Evaluation of northern right > Fujiwara M, Caswell H (2001) Demography of the endan- whale ship strike reduction measures in the Great South gered North Atlantic right whale. Nature 414:537–541 Channel of Massachusetts. NOAA Tech Memo NMFS- Gaskin DE (1983) The marine mammal community. In: NE-202, p 1–12 Thomas MLH (ed) Marine and coastal systems of the > Michaud J, Taggart CT (2007) Lipid and gross energy content Quoddy region, New Brunswick. Can Spec Pub Fish of North Atlantic right whale food, Calanus finmarchicus, Aquat Sci 64, p 245–268 in the Bay of Fundy. Endang Species Res 3:77–94 Gaskin DE (1987) Updated status of the right whale, Eubal- Moore MJ, McLellan WA, Daoust PY, Bonde RK, Knowlton aena glacialis, in Canada. Can Field Nat 101:295–309 AR (2007) Right whale mortality: a message from the IMO (International Maritime Organization) (2003) New and dead to the living. In: Kraus SD, Rolland RM (eds) The amended traffic separation schemes. REF T2/2.07, COL- urban whale: North Atlantic right whales at the cross- REG.2/Circ.52. IMO, London roads. Harvard University Press, Cambridge, MA, IMO (International Maritime Organization) (2007a) Routing p 358–379 of ships, ship reporting, and related matters. NAV 53/3/13, NARWC (North Atlantic Right Whale Consortium) (2005) IMO, London North Atlantic Right Whale Consortium sightings data- IMO (International Maritime Organization) (2007b) Routing base, 02 March 2005. New England Aquarium, Boston, measures other than Traffic Separation Schemes. Ref. T2- MA OSS/2.7, SN.1/Circ.263, IMO, London NOAA (National Oceanic and Atmospheric Administration) IWC (International Whaling Commission) (2001) Report of (2006) Proposed rule to implement speed restrictions to the workshop on status and trends of western North reduce the threat of ship collisions with North Atlantic Atlantic right whales. J Cetacean Res Manag 2(Spec right whales. Fed Regist 71:36299–36319 Issue):61–87 > Nowacek DP, Johnson MP, Tyack PL (2004) North Atlantic Kenney RD (2001) The North Atlantic Right Whale Consor- right whales (Eubalaena glacialis) ignore ships but tium databases. Maritimes 43:3–5 respond to alerting stimuli. Proc R Soc Lond Ser B Biol Sci Kenney RD, Mayo CA, Winn HE (2001) Migration and forag- 271:227–231 ing strategies at varying spatial scales in western North > Panigada S, Pesante G, Zanardelli M, Capoulade F, Gannier Atlantic right whales: a review of hypotheses. J Cetacean A, Weinrich MT (2006) Mediterranean fin whales at risk Res Manag 2(Spec Iss):251–260 from fatal ship strikes. Mar Pollut Bull 52:1287–1298

015

Vanderlaan et al.: Reducing vessel–whale lethal encounter risk 297

> Rosenbaum HC, Brownell RL, Brown MW, Schaeff C and Marine mammal sensory systems. Plenum, New York, others (2000) World-wide genetic differentiation of Eubal- p 727– 739 aena: questioning the number of right whale species. Mol > Trippel EA, Strong MB, Terhune JM, Conway JD (1999) Miti- Ecol 9:1793–1802 gation of harbour porpoise (Phocoena phocoena) by-catch Scott GP, Gilbert JR (1982) Problems and progress in the US in the gillnet fishery in the lower Bay of Fundy. Can J Fish BLM-sponsored CETAP surveys. Rep Int Whaling Comm Aquat Sci 56:113–123 32:587–600 > Vanderlaan ASM, Taggart CT (2007) Vessel collisions with Stone GS, Kraus SD, Prescott JH, Hazard KW (1988) Signifi- whales: the probability of lethal injury based on vessel cant aggregations of the endangered right whale, Eubal- speed. Mar Mamm Sci 23:144–156 aena glacialis, on the continental shelf of Nova Scotia. Can > Vanderlaan ASM, Hay AE, Taggart CT (2003) Characteri- Field Nat 102:471–474 zation of North Atlantic right-whale (Eubalaena glacialis) Todd S, Lien J, Verhulst A (1992) Orientation of hump- sounds in the Bay of Fundy. IEEE J Ocean Eng 28:164–173 back whales (Megaptera novaeangliae) and minke > Waldick RC, Kraus S, Brown M, White BN (2002) Evaluating whales (Balaenoptera acutorostrata) to acoustic alarm the effects of historic bottleneck events: an assessment of devices designed to reduce entrapment in fishing microsatellite variability in the endangered, North gear. In: Thomas JA, Kastelein RA, Supin AY (eds) Atlantic right whale. Mol Ecol 11:2241–2249

Editorial responsibility: Wayne Linklater, Submitted: November 1, 2007; Accepted: January 31, 2008 Wellington, New Zealand Proofs received from author(s): March 21, 2008

016 Ship noise extends to frequencies used for echolocation by endangered killer whales

Scott Veirs1, Val Veirs2 and Jason D. Wood3 1 Beam Reach Marine Science and Sustainability School, Seattle, WA, United States 2 Department of Physics, Colorado College, Colorado Springs, CO, United States 3 SMRU Consulting, Friday Harbor, WA, United States

ABSTRACT Combining calibrated hydrophone measurements with vessel location data from the Automatic Identification System, we estimate underwater sound pressure levels for 1,582 unique ships that transited the core critical habitat of the endangered Southern Resident killer whales during 28 months between March, 2011, and October, 2013. Median received spectrum levels of noise from 2,809 isolated transits are elevated relative to median background levels not only at low frequencies (20–30 dB re 1 µPa2/Hz from 100 to 1,000 Hz), but also at high frequencies (5–13 dB from 10,000 to 96,000 Hz). Thus, noise received from ships at ranges less than 3 km extends to frequencies used by odontocetes. Broadband received levels (11.5–40,000 Hz) near the shoreline in Haro Strait (WA, USA) for the entire ship population were 110 ± 7 dB re 1 µPa on average. Assuming near-spherical spreading based on a transmission loss experiment we compute mean broadband source levels for the ship population of 173 ± 7 dB re 1 µPa 1 m without accounting for frequency-dependent absorption. Mean ship speed was 7.3 ± 2.0 m/s (14.1 ± 3.9 knots). Most ship classes show a linear relationship between source level and speed with a slope near +2 dB per m/s (+1 dB/knot). Spectrum, 1/12-octave, and 1/3-octave source levels for the whole population have median values that are comparable to previous measurements and models at most frequencies, but for select studies may be relatively low below 200 Hz and high above 20,000 Hz. Median source spectrum levels peak near 50 Hz for all Submitted 29 September 2015 12 ship classes, have a maximum of 159 dB re 1 µPa2/Hz @ 1 m for container ships, Accepted 13 January 2016 and vary between classes. Below 200 Hz, the class-specific median spectrum levels Published 2 February 2016 bifurcate with large commercial ships grouping as higher power noise sources. Within Corresponding author all ship classes spectrum levels vary more at low frequencies than at high frequencies, Scott Veirs, [email protected] and the degree of variability is almost halved for classes that have smaller speed Academic editor standard deviations. This is the first study to present source spectra for populations Magnus Johnson of different ship classes operating in coastal habitats, including at higher frequencies Additional Information and used by killer whales for both communication and echolocation. Declarations can be found on page 29

DOI 10.7717/peerj.1657 Subjects Conservation Biology, Environmental Sciences, Marine Biology, Science Policy, Coupled Copyright Natural and Human Systems 2016 Veirs et al. Keywords Noise, Hydrophone, Killer whale, Orca, Odontocete, Marine mammal, Ship, Pollution, Acoustics, Bioacoustics Distributed under Creative Commons CC-BY 4.0

OPEN ACCESS

How to cite this article Veirs et al. (2016), Ship noise extends to frequencies used for echolocation by endangered killer whales. PeerJ 4:e1657; DOI 10.7717/peerj.1657 017 INTRODUCTION Commercial ships radiate noise underwater with peak spectral power at 20–200 Hz (Ross, 1976). Ship noise is generated primarily from propeller cavitation, propeller singing, and propulsion or other reciprocating machinery (Richardson et al., 1995; Wales & Heitmeyer, 2002; Hildebrand, 2009). The dominant noise source is usually propeller cavitation which has peak power near 50–150 Hz (at blade rates and their harmonics), but also radiates broadband power at higher frequencies, at least up to 100,000 Hz (Ross, 1976; Gray & Greeley, 1980; Arveson & Vendittis, 2000). While propeller singing is caused by blades resonating at vortex shedding frequencies and emits strong tones between 100 and 1,000 Hz, propulsion noise is caused by shafts, gears, engines, and other machinery and has peak power below 50 Hz (Richardson et al., 1995). Overall, larger vessels generate more noise at low frequencies (<1,000 Hz) because of their relatively high power, deep draft, and slower-turning (< 250 rpm) engines and propellers (Richardson et al., 1995). This low-frequency energy from ships is the principal source of ambient noise within the deep ocean from approximately 5–1,000 Hz (Wenz, 1962; Urick, 1983; National Research Council et al., 2003). Growth of the global shipping fleet and possibly the average size of ships has raised deep-ocean ambient noise levels in low-frequency bands near 40 Hz by up to 20 dB relative to pre-industrial conditions (Hildebrand, 2009) and 8–10 dB since the 1960s (Andrew et al., 2002; McDonald, Hildebrand & Wiggins, 2006). As these ships enter shallow waters and traverse the estuarine habitat typically occupied by major ports, the noise they radiate may impact coastal marine life. Since many marine mammals rely on sound to find prey, moderate social interactions, and facilitate mating (Tyack, 2008), noise from anthropogenic sound sources like ships can interfere with these functions, but only if the noise spectrum overlaps with the hearing sensitivity of the marine mammal (Southall et al., 2007; Clark et al., 2009; Hatch et al., 2012). Hearing sensitivity isn’t yet characterized in Mysticetes (baleen whales), but based on their signals they are likely most sensitive at frequencies 10–10,000 Hz and therefore constitute a low-frequency functional hearing group (Southall et al., 2007). They typically emit signals with fundamental frequencies well below 1,000 Hz (Cerchio, Jacobsen & Norris, 2001; Au et al., 2006; Munger et al., 2008) although non-song humpback signals have peak power near 800 and 1,700 Hz (Stimpert, 2010) and humpback song harmonics extend up to 24,000 Hz (Au et al., 2006). The frequency overlap of peak power in ship noise and baleen whale signals (and inferred maximum hearing sensitivity) is verified by observed behavioral and physiological responses of mysticetes to ship noise. As examples, the probability of detecting a blue whale D call increases in ship noise, suggesting a Lombard effect (Melcon et al., 2012) and Rolland et al. (2012) found decreased stress levels in North Atlantic right whales when ship noise was absent. The potential impacts of ship noise can be assessed more confidently in Odontocetes (toothed whales) because they constitute mid-frequency or high-frequency functional hearing groups (Southall et al., 2007) in which auditory response curves have been obtained for many species. These curves show maximum auditory sensitivity near the

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 2/35 018 frequencies where toothed whale signals have peak power (Mooney, Yamato & Branstetter, 2012; Tougaard, Wright & Madsen, 2014)—at about 1,000–20,000 Hz for social sounds and 10,000–100,000 Hz or higher for echolocation. Southern Resident killer whales (SRKWs) represent an endangered toothed whale species that inhabits an urban estuary in which shipping traffic is common and is very well characterized bioacoustically. Their auditory sensitivity, extrapolated from captive killer whales (Hall & Johnson, 1972; Szymanski et al., 1999), peaks at 15,000–20,000 Hz—a frequency range that overlaps with the upper range of their vocalizations and the lower range of their echolocation clicks. SRKW calls have fundamental frequencies at 100– 6,000 Hz with harmonics extending up to 30,000 Hz (Ford, 1987). Their echolocation clicks are likely similar to those of salmon-eating northern resident killer whales which have a 40,000 Hz bandwidth and a mean center frequency of 50,000 Hz (Au et al., 2004). SRKWs whistle between 2,000 and 16,000 Hz (Riesch, Ford & Thomsen, 2006) with a mean dominant frequency of 8,300 Hz (Thomsen, Franck & Ford, 2000). Behavioral responses to boat (as opposed to ship) noise have been documented in toothed whales, including SRKWs. For example, bottlenose dolphins whistle (at 4,000– 20,000 Hz) less when exposed to boat noise at 500–12,000 Hz (Buckstaff, 2004) and Indo- Pacific bottlenose dolphins lower their 5,000–10,000 Hz whistle frequencies when noise is increased by boats in a band from 5,000 to 18,000 Hz (Morisaka et al., 2005). For every 1 dB increase in broadband underwater noise (1,000–40,000 Hz) associated with nearby boats, SRKWs compensate by increasing the amplitude of their most common call by 1 dB (Holt et al., 2009). While the frequencies used by toothed whales are well above the peak power frequencies of ships, multiple lines of evidence suggest that ship noise spectra extend or should be expected to extend to higher frequencies. Laboratory experiments with cavitation and previous studies of submarines, torpedoes, and ships indicate that ship noise may extend as high as 160,000 Hz at the source. Experiments confirm that cavitation generates high frequency noise up to at least 100,000 Hz (Wenz, 1962). Cavitation noise from spinning rods and water jets has spectral power that rises through low frequencies at a rate of 40 dB/decade to a peak near 1,000 Hz and thereafter descends at −20 dB/decade (Mellen, 1954; Jorgensen, 1961). Noise from foil cavitation also has peak spectral power at 1,000 Hz, as well as a secondary peak at 31,000 Hz (Blake, Wolpert & Geib, 1977). In the vicinity of the higher peak, 1/3-octave levels increase about 10 dB upon cavitation inception (Blake, Wolpert & Geib, 1977). World War II studies of ship noise, particularly measurements of thousands of transits of hundreds of ships of all types, identified propeller cavitation as the dominant source of noise radiated by ships, including at high frequencies (Dow, Emling & Knudsen, 1945). In reviewing these studies Ross (1976) and Urick (1983) noted that increases of >40 dB in the 10,000–30,000 Hz band were diagnostic of cavitation inception on accelerating twin- screw submarines and Urick (1983) attributed a 1 dB/knot (2 dB per m/s) rise in torpedo spectrum levels from 10,000 to 75,000 Hz to propeller cavitation. More recently, cavitation has been implicated in ship noise measurements made at close range (<1,000 m) which show levels between 1,000 and 60,000 Hz that not

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 3/35 019 only are significantly above background levels, but also rise with increased ship speed faster than at lower frequencies (Arveson & Vendittis, 2000; Kipple, 2002; Hermannsen et al., 2014). Even when portions of the high-frequency energy are excluded, broadband source levels of cavitating propellers are high. Erbe & Farmer (2000) reported median broadband (100–20,000 Hz) source levels for an icebreaker with a cavitating propeller of 197 dB re 1 µPa @ 1 m. In the open ocean or on the outer continental shelf far from shipping lanes high- frequency noise radiated by a ship will be absorbed within about 10 km (Erbe & Farmer, 2000), often before reaching a species of concern. In urban estuaries, however, marine mammals are exposed to noise from ships at ranges of 1–10 km routinely, and less than 100 m occasionally. For example, SRKWs frequently transit Haro Strait within 10–300 m of the shoreline at Lime Kiln Point where they are about 2 km from the center of the northbound (nearest) shipping lane (Fig. 1). Since the absorption rate is only about 3 dB/km at 20,000 Hz, compared to 30 dB/km at 100,000 Hz (Francois & Garrison, 1982), ship noise near 20,000 Hz (where SRKWs are most sensitive) in such close quarters may retain the potential to mask echolocation clicks, as well as other high-frequency signals. In an environment where SRKWs may already be food-stressed (Ayres et al., 2012) due to reduced populations of their primary prey—Chinook salmon (Hanson et al., 2010)— echolocation masking could have grave population-level consequences. The potential impacts of ship noise on foraging efficiency may be compounded by simultaneous masking of communication calls, some of which may help coordinate foraging or prey sharing (Ford & Ellis, 2006). One case study has suggested that ship noise may reduce foraging efficiency by 50% in Curvier Beaked whales (Aguilar Soto et al., 2006). Motivated by the possible impacts of ship noise on odontocetes and the scarcity of ship noise measurements made at close range over the full range of frequencies used by SRKWs, we endeavor to estimate source spectrum levels up to 40,000 Hz for a wide variety of ships from measurements made at a range of less than a few kilometers. Our primary objective is to characterize ship noise at higher frequencies, specifically those important to killer whales. A secondary objective is to compare our results with previous studies in order to understand consistencies and possible biases in field measurements of ship noise.

METHODS Our study site is an area of the inland waters of Washington State and British Columbia known as the Salish Sea. This urban estuary hosts the commercial shipping ports of Vancouver, Seattle, and Tacoma (see Fig. 1). Shipping traffic primarily associated with Vancouver—about 20 large (>65 feet 1All decibels here are referred to 1 µPa and or 19.8 m) vessels per day (Veirs & Veirs, 2006)— transits Haro Strait, the core of the source levels to a distance of 1 m. After their first usage, the units of broadband summertime habitat of the SRKWs (Hauser et al., 2007). Each ship typically raises sound and spectrum level decibels are generally pressure levels1 near the shoreline about 20 dB re 1 µPa (RMS, 100–15,000 Hz) above suppressed. background levels to about 115 dB for approximately 20 min/transit (Veirs & Veirs, 2006). We define ships as all vessels with overall length (LOA) greater than 65 feet (19.8 m); the remaining, shorter vessels (boats) are not characterized in this study.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 4/35 020 Figure 1 Study site map. Inset regional map shows the study area (black rectangle) and shipping lanes (in red) leading to the major ports of the Salish Sea. The 240◦ bearing (gray arrow) extends from the Lime Kiln hydrophone (gray circle) through the northbound shipping lane. Bathymetric contours (50 m) show that Haro Strait is a steep-sided 200–300 m-deep channel. Sound projection locations (black dots) are sites used for the transmission loss experiment.

We measured underwater noise radiated by these ships, collecting data continuously during 28 months between March 7, 2011, and October 10, 2013, except for occasional 1–2 day interruptions caused by power outages. About 3.5 months of data were excised due to systematic noise caused during equipment repairs made between July 22, 2011, and November 9, 2011. Consequently, we sampled every month of the year at least twice. Study site We deployed a calibrated hydrophone 50 m offshore of the lighthouse at Lime Kiln State Park in which The Whale Museum and Beam Reach maintain an acoustic observatory as part of the Salish Sea Hydrophone Network (orcasound.net). Midway along the west side of San Juan Island, Lime Kiln lighthouse sits on a point near the center of the

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 5/35 021 Figure 2 Comparison of source levels from different studies for various classes of ships. Broadband source level (SL) statistics for each ship class juxtaposed with results from recent studies of comparable classes. Bold horizontal lines are medians; gray box hinges are 25% and 75% quantiles; gray whiskers extend to the value that is most distant from the hinge but within 1.5 times the inter-quartile range (distance between the 25% and 75% quantiles); red dots are mean values from Table 2. Each encircled letter B represents a mean from Bassett et al. (2012); blue vertical bars represent means from McKenna et al. (2012) with the container ship estimate of McKenna, Wiggins & Hildebrand (2013) labeled McKenna; black vertical bars represent estimates from Kipple (2002) and Arveson & Vendittis (2000).

summertime habitat of the SRKWs (Fig. 1). While the killer whales sometimes swim directly over the hydrophone location, they more typically transit the site 100–300 m offshore where received levels of noise from the shipping lanes would be somewhat higher than those recorded in this study. The hydrophone was secured to a PVC pipe projecting vertically from a cement- filled tire resulting in a position 1 m above the bottom at a depth of 8 m (below mean lower low water). A cable protected by irrigation pipe secured in the inter- and sub-tidal zones brought the signal to recording hardware within the lighthouse and also housed a saltwater ground wire that helped reduce system noise. The local bathymetry on a transect perpendicular to the shoreline (240◦ bearing) and running from the hydrophone to the northbound shipping lane descends to deep (>200 m) water within 300 m of the shoreline. The nearshore region (<150 m from shore) has a substrate of boulders and gravel covered with marine vegetation and descends at a slope of about 20◦. Further from shore the bottom descends at a slope of about 45◦. Relative to the northbound shipping lane the hydrophone position is 1.3 km from the eastern edge, 2.25 km from the center of the lane, and about 3.7 km from the center of the traffic separation zone. A histogram of the range to all ships in our database shows peaks at 2.3 and 5.0 km, corresponding with the middle of the north- and south-bound lanes, respectively.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 6/35 022 Data acquisition We made audio recordings of the signal from a Reson TC4032 hydrophone installed with a differential output (sensitivity of −164 ± 3 dB re 1 V/µPa from 5 to 125,000 Hz) that was amplified and then digitized by a MOTU Traveller sampling at 192,000 Hz with 16 bits per sample. The maximum signal that could be recorded without clipping was 140 dB. A Windows XP computer analyzed and archived the recorded signal. We calibrated the recording system with the analog output of an Interoceans 902 (acoustic listening calibration system) while a ship was passing the lighthouse, thereby converting the samples to decibels (dB) referenced to 1 µPa (hereafter dB re 1 µPa). This procedure was carried out occasionally to check and make minor changes in the Reson calibration constant during the 28 month study period. A Python program analyzed the digitized hydrophone signal. The program continuously computed running 2-second mean square voltage levels. Each hour the program archived the 2-second recordings that yielded the minimum and maximum averages. We used the minimum files to determine background noise levels. Generally, all commercial ships over 300 tons are required to use the Automatic Identification System (AIS) to broadcast navigational data via VHF radio. The AIS carriage requirements of the US Coast Guard (33 CFR 164.46) and Canada within a vessel traffic service area like Haro Strait mean that some fishing and passenger vessels may be underrepresented in our data set. Each AIS-equipped ship transmits at least its identification number, location, course, and speed a few times each minute. The typical range over which these transmissions are detected is 45 km. The Python program scanned the binary output of an AIS receiver (Comar Systems AIS-2-USB) located in the lighthouse. For each transmission received, the location of the ship was used to calculate its range (R) from the hydrophone. When R was less than 4 nautical miles (7.4 km), the program recorded the broadband received level every 0.5 nautical mile (926 m) as the ship approached and departed. When the ship crossed a line perpendicular to shore (at an azimuth angle of 240◦ true, see Fig. 1), the Python program stored a 30-second WAV file, the date and time, and the decoded ship metadata (ship ID number, range, speed over the ground (SOG), and course over the ground). Given the orientation of the northbound shipping lane, this procedure made it likely that we recorded the starboard beam aspect noise levels of each isolated ship near the closest point of approach. Finally, the program calculated the calibrated broadband received level using the Reson calibration constant and the RMS amplitude of the 30-second file. To maximize the detection of any high-frequency signal generated by passing ships, and to reduce the spatial extent of our transmission loss experiment, we elected to compute source levels for only the closer, northbound portion of the traffic in Haro Strait. Southbound traffic was recorded, counted, and archived, but is not included in this analysis. For the northbound traffic presented herein, the mean and standard deviation of R is 2.30 ± 0.39 km, and the minimum and maximum R are 0.95 km and 3.65 km, respectively.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 7/35 023 Data analysis Isolation and identiﬁcation Archived WAV files and associated metadata were analyzed with a C++ program developed in the platform-independent Qt environment (qt-project.org). To measure the noise radiated by an individual ship, rather than multiple ships, the program used the AIS data to detect acoustically-isolated ships. A ship was deemed isolated if the previous and subsequent ships were at least 6 nautical miles (11.1 km) away from the hydrophone when the WAV file was recorded. It is only at closer range that human listeners can detect ship noise above ambient levels. For each isolated ship, the program used the ship’s identification (Maritime Mobile Service Identity, or MMSI) number to look up details about the ship from online web sites such as the Marine Traffic network (www.marinetraffic.com). These metadata, saved in a MySQL database, include (when available): MMSI, ship name, ship type, year built, length, breadth, dead weight, maximum and mean speed, flag, call sign, IMO, draft, maximum draft, and photographs. We simplified 41 ship type categories returned from online queries into 12 general ship classes: bulk carrier (includes ore carriers); container; tug (includes multi-purpose offshore, pusher tug, and tender); cargo (includes other cargo, heavy lift, wood chip carrier); vehicle carrier (includes all roll-on roll-offs); tanker (includes crude oil, oil product, oil/chemical, chemical, and product tankers); military (includes Coast Guard, search and rescue); fishing (includes fish carrier, factory, fishing, fishing vessel, and trawler); passenger (includes cruise ships and ferries); miscellaneous (includes cable layer, reserved, unspecified, and well-stimulation); pleasure craft (includes sailing vessels, motor yacht, and yachts); and research.

Received levels From each isolated ship’s WAV file the RMS power spectral density (PSD) was calculated using a Fast Fourier Transform averaged over the 30-second duration of the file (Nyquist frequency of 96,000 Hz; 16,384 (214) sample overlapping Bartlett window). The bandwidth of each of the 8,192 frequency bins was 11.5 Hz. These RMS PSD (per Hz) values were calibrated by requiring that the integral of the PSD equal the calibrated broadband level associated with each WAV file. The resulting power spectral densities we call the total received spectrum levels. The total received spectrum level is a composite of the power that originated from the ship and the power associated with the background noise at the time of the ship passage. To enable estimation of the background level at the time of ship passage we continuously observed 2-second sound samples, saving the lowest power 2-second sample every hour. The subtraction of the estimated background received level (RLB) from the total received spectrum level (RLT ) to determine the received spectrum level associated with the ship (RLS) is based on the fact that when two or more waves pass at once, the pressure on the hydrophone (P) is the sum of the instantaneous pressure from each wave. The power that we calculate is proportional to the square of the pressure on the hydrophone

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 8/35 024 and is represented in decibels. These relationships apply both for the power at individual frequencies (PSD) and the total power (PwrT ) integrated over all frequencies. Following the nomenclature of Erbe (2011),

2 PwrT (t) = k(PS(t)+PB(t)) , (1)

where k is a constant dependent on the construction of the hydrophone and t is time. Averaging over the 30 s of each WAV file, we assume that the pressure due to the ship at each moment in time is not correlated with the pressure due to other (background) noise sources. Thus, the power received from the ship is the average total power minus the average background power:

hPwrSi = hPwrT i−hPwrBi. (2)

We estimate PwrB for each passing ship as the average of the power in two samples— the quietest 2-second sample from the hour before the ship is recorded, and quietest from the hour after the ship passage. On occasion during daylight hours, ship recordings contain noise from vessels unequipped with AIS (usually recreational motorboats and occasionally larger vessels operating without AIS). This contamination is limited to the 50, 75, and 95% quantiles above 20,000 Hz, has peak spectrum levels near 50,000 Hz—a frequency commonly used for depth sounders—and is rare, but we have nevertheless reduced it via a 2-step statistical process. Since it is very rare to have motorboat noise overlapping with ship passage at night, we first determined the 95% quantile of each received spectrum level across all vessels recorded at night (hour of day greater than 19:00 or less than 07:00) and used it as a threshold above which contamination by boat noise may have occurred. Then we re-processed all ship transits, removing any data points for which the threshold was exceeded. Any recording in which at least 100 of the 8,192 spectral received levels were above threshold was omitted from further statistical analysis. Through this robust statistics process, about 15% of transits were omitted, resulting in no difference between the ship population quantiles for ships that pass during the day versus the night. A sensitivity analysis shows that the process did not affect the 5%–75% quantiles and that the 95% quantile was reduced by less than 2 dB—and only above about 20,000 Hz. The high frequency peaks seen in the 95% quantile in Fig. 3 become sharper as the threshold is increased or the total number of vessels analyzed is decreased. Finally, we report received levels (RL) in decibels relative to a reference pressure of 1 µPascal and estimate ship received levels as:

= RLT /10 − RLB/10 RLS 10log10 10 10 . (3)

Often RLT is much higher than RLB at all frequencies. In such cases, subtraction of the background has little effect on RLS. But for many ships RLT is close to RLB, at least at some frequencies. Therefore, we subtract the estimated background from the RLT at all frequencies for every isolated ship, yielding the received spectrum level of ship noise, RLS.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 9/35 025 Figure 3 Quantiles of received spectra for background and ship noise. Quantiles (5, 25, 50, 75, & 95%) of background spectrum level (RLB, dashed blue lines) and total received spectrum level for the entire ship population (RLT , solid black lines).

We cannot determine RLS if the associated RLT is not greater than RLB. Hence we require that RLT at any given frequency must exceed a threshold of three times the background spectrum level at that frequency. We choose this factor (4.8 dB) by examining the statistics of typical ship and background recordings to assure that noise is unlikely to be taken as signal. We refrain from reporting ship source spectra above 40,000 Hz because the sample size in bands above this frequency falls below about 10% of the mean sample size in lower frequency bands. Furthermore, to calculate broadband source levels with or without absorption we integrate the spectrum levels only up to this 40,000 Hz upper limit. Prior to the background subtraction, our data commonly contained narrow-band noise peaks near 25, 38, 43, 50 and 69 kHz in many of the background and total received spectrum level quantiles (Fig. 3). Unknown sources of transient systematic noise (most commonly near 77 kHz), typically lasted only a few days. Because these noise sources are narrow or brief, they contain little power. Also, since they occur in both the received level and background data, they tend to be removed through background subtraction, and therefore do not significantly contaminate the estimated source levels (Fig. 4). One exception is the peak near 25 kHz—likely associated with the Jim Creek Naval Radio Station (transmitting at 24.8 kHz)—which persists in many source level spectra, probably indicating that the submarine communications are intermittent, at times occurring during a ship passage but not during the corresponding background measurements.

Transmission loss experiment

To estimate the source spectrum level of isolated ships from RLS we measured the transmission loss along the 240◦ true bearing line from the near-shore hydrophone at Lime

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 10/35 026 Figure 4 Ship noise source spectrum, 1/12-, and 1/3-octave levels. Source level (SL) spectra of the entire ship population in 1 Hz (solid), 1/12-octave (dashed), and 1/3-octave bands (dotted). Black curves are medians without absorption; red curves are medians with absorption. For the spectrum levels, we delin- eate 25 and 75% quantiles in lighter tones. Levels with absorption start to increase rapidly above 15,000– 20,000 Hz for both the 1/12- and 1/3-octave bands.

Kiln into the northbound shipping lane (Fig. 1). The transmission loss is a combination of geometric spreading and frequency-dependent absorption. While Haro Strait has less distinct winter and summer sound speed profiles than other parts of the Salish Sea due to vertical mixing by tidal flow over bounding sills, to average out any seasonal effects we conducted our transmission loss experiment in the spring. We determined the geometric spreading via a field experiment conducted in March, 2014, from a 10 m catamaran. We projected a sequence of 2-second tones (Table 1) using a Lubell 9816 underwater speaker lowered in a bifilar harness from the bows and attached to a power amplifier and a digital sound player. During each tone sequence, we noted the location of the projector on the sailboat’s GPS and measured the projected sound level with the Interoceans 402 hydrophone, having positioned its calibrated hydrophone near the stern, about 10 m from the projector. We oriented the projection system toward the lighthouse as we played each sequence at the following distances from the projector to the Lime Kiln hydrophone: 290; 1,035; 1,446; and 2,893 m. This study focuses on determining the source levels of ships that are northbound at Lime Kiln lighthouse. By limiting our analysis to northbound vessels we reduce the difficulty of determining accurate transmission loss by limiting the variation in range of the targets. Furthermore, our underwater speaker used to measure transmission loss did not have sufficient power especially at high frequencies (near 20,000 Hz) to provide detectable signals at ranges much larger than the 2,893 m range that brackets the more distant edge of the north bound traffic lane.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 11/35 027 Table 1 Results of the transmission loss experiment. For each projected frequency, the geometric spreading rate (TL) is near-spherical, with an average slope of −18.6±0.4 dB/decade.

Frequency (Hz) TL (dB/decade) Coefficient of determination 630 −18.85 0.926 1,260 −18.08 0.991 2,510 −18.99 0.986 5,000 −18.24 0.964 10,000 −18.37 0.974 15,000 −19.09 0.987 20,100 −18.67 0.971

We analyzed the spreading of the test tones by measuring the calibrated RMS level received at the Lime Kiln hydrophone for each tone at each distance. The received signal was determined by subtracting the calibrated background level from the received level of the corresponding tone (Eq. (3)). To determine the geometric spreading contribution to transmission loss, we added to the received signal levels the amount of absorption expected for each frequency and range (straight line path, R). Following Francois & Garrison (1982) we used R to calculate the absorption loss at each frequency. For our highest test tone frequencies and range, accounting for absorption added from 2 dB (at 10,000 Hz) to 8.6 dB (at 20,000 Hz) back into the received signal levels. We used linear regression to model the absorption-corrected received signal levels as a function of the base 10 logarithm of the range from receiver to source in meters separately for each of our test tones. The slopes and goodness of fit are shown in Table 1. Since these slopes are not correlated with the frequency (correlation coefficient of 0.003), we average them and use the resulting near-spherical geometric spreading coefficient − ± (transmission loss coefficient, TL) of 18.6 0.4 dB/decade in log10(R) to represent geometric spreading out to a distance of about 3 km. Also, as these slopes vary little over a factor of 30 in frequency, we assume that we can use this mean slope to extrapolate down from 630 Hz to our 20 Hz lower frequency cutoff and up from 20,000 Hz to our 96,000 Hz upper frequency Nyquist cutoff.

Source levels We calculate source spectrum levels of ship noise without absorption (SL) in Eq. (4) and then with absorption (SLa) in Eq. (5), determining α from Francois & Garrison (1982).

= + SL RLS 18.6log10(R) (4) = + + SLa RLS 18.6log10(R) α(f )R. (5)

We integrate the source spectrum levels from 11.5 Hz up to 40,000 Hz to compute broadband source levels (SL)(Table 2). We also integrate the source spectrum levels over both 1/3-octave and 1/12-octave bands with band centers determined by f (i) = i fo2 N where i is an integer and N is the number of partitions of each octave. This is

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 12/35 028 Table 2 Ship population statistics and mean broadband sound pressure levels (20–40,000 Hz). Though abbreviated in the table as dB, the units of the received signal levels (RLS) are dB re 1 µPa and source levels have units of dB re 1 µPa @ 1 m. Variability is reported as a standard deviation of the mean, and speed over ground (SOG) is provided in m/s and knots.

Ship class Isolated transits % of total Unique ships RLS dB SL dB SOG m/s SOG knots All classes combined 2,809 1,582 110 ± 7 173 ± 7 7.3 ± 2.0 14.1 ± 3.9 Bulk carrier 965 34.3 734 111 ± 6 173 ± 5 7.0 ± 0.8 13.7 ± 1.5 Container 529 18.8 207 116 ± 4 178 ± 4 9.9 ± 1.0 19.2 ± 1.9 Tug 337 12.0 85 108 ± 5 170 ± 5 4.2 ± 1.2 8.2 ± 2.3 Cargo 307 10.9 206 113 ± 5 175 ± 5 7.4 ± 1.2 14.4 ± 2.3 Vehicle carrier 187 6.6 111 113 ± 3 176 ± 3 8.7 ± 1.0 16.9 ± 1.8 Tanker 148 5.3 101 111 ± 5 174 ± 4 7.1 ± 0.7 13.8 ± 1.4 Military 113 4.0 19 99 ± 10 161 ± 10 5.7 ± 1.6 11.1 ± 3.1 Fishing 65 2.3 28 102 ± 9 164 ± 9 4.7 ± 1.1 9.1 ± 2.2 Passenger 49 1.7 31 104 ± 8 166 ± 8 7.4 ± 2.3 14.4 ± 4.5 Miscellaneous 41 1.4 21 101 ± 9 163 ± 9 5.8 ± 3.0 11.2 ± 5.8 Pleasure craft 41 1.5 35 97 ± 10 159 ± 9 6.4 ± 2.5 12.4 ± 4.9 Research 14 0.5 4 104 ± 6 167 ± 5 5.7 ± 0.9 11.1 ± 1.8

both consistent with ISO center frequencies (ISO 266) and allows comparison with the proposed annual mean noise thresholds at 63 and 125 Hz Tasker et al. (2010); Merchant et al. (2014). Finally, when plotting quantiles of levels we exclude the lowest frequency bin (11.5 Hz) because for some classes an insufficient number of ships passed the 4.8 dB re 1 µPa signal-noise threshold to estimate the 5% and 95% quantiles. To facilitate comparison with past studies we generally present ship source spectrum levels as SL. However, due to the presence of high-frequency ship noise in our recordings and its potential impact on marine life exposed at close range, we also present absorption- corrected spectral power levels (SLa) for the whole ship population.

RESULTS AND DISCUSSION Ship traﬃc patterns Combining all ship classes over the entire study, our data set describes 1,582 unique vessels that made a total of 2,809 isolated, northbound transits of the shipping lanes in Haro Strait (Table 2). The 2,809 isolated transits sample 17.1% of the total transits through Haro Strait (16,357, northbound and southbound) logged by our AIS system during the study period. Of 7,671 total northbound transits, 36% were sampled, suggesting that about 2/3 of the traffic in Haro Strait is not isolated. Dividing the total transits by the 850 day study period shows that the average daily ship traffic is 19.5 ships/day. This amount of traffic is comparable to previous estimates for Haro Strait: about 20 ships/day (Veirs & Veirs, 2006) and about 1 ship/hour (Erbe, MacGillivray & Williams, 2012). About 1/3 of the isolated transits are bulk carriers and about 1/5 are container ships. The next 4 most prevalent ship classes—tugs, cargo ships, vehicle carriers, and tankers— constitute another 1/3 of the isolated transits. Of the remaining less-prevalent ship classes,

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 13/35 029 we sample military ships 113 times (19 unique vessels), and other ship classes 18–65 times. Together, bulk carriers and container ships comprise more than half (53%) of the isolated shipping traffic in Haro Strait. About 3/4 of isolated bulk carrier transits are unique vessels, in contrast to container ships which are unique only about 40% of the time. This may indicate that the global bulk carrier fleet is larger than the container fleet, or that shipping economics or logistics limit the diversity of container ships transiting Haro Strait. For example, container ships may ply routes that are more fixed, and therefore repeat transits through Haro Strait more frequently than bulk carriers. Those ship classes that have many isolated transits by a small number of unique ships offer us opportunities to study variability of noise from individual ships. Military vessels, a category with 19 unique ships sampled on 113 isolated transits, have about 7 isolated transits per unique ship, while tugs and research vessels have about 4 and container ships have about 3. Broadband levels Received levels

Broadband population mean received levels (RLS, Table 2) vary between ship classes from a low of 97 dB (pleasure craft) to a high of 116 dB (container ships). Combining all classes, RLS is 110 ± 7 dB which is 19 dB re 1 µPa above the mean background level (RLB) of 91 ± 4 dB. These levels are comparable to anthropogenic and background received levels noted in previous studies at similar distances to shipping lanes and over similar frequency ranges (Veirs & Veirs, 2006; McKenna et al., 2012). While our RLS from ships 0.95–3.65 km away is 10–22 dB lower than the 121–133 dB reported by Bassett et al. (2012), only about 2 dB of this difference can be explained by the shorter distances to their ships (0.58–2.82 km).

Source levels (SL) The mean broadband source level (SL, Table 2) for all ship classes combined is 173 ± 7 dB re 1 µPa @ 1 m. Comparing between ship classes, container ships have the highest SL at 178 dB. Other classes with SL ≥ 173 dB include vehicle carriers, cargo ships, tankers, and bulk carriers. Tugs, research, and passenger vessels (primarily cruise ships, as there are no nearby ferry routes) have SL of 166–170 dB, while the remaining vessel classes have SL from 159–164 dB. This range of SL across classes (159–178 dB) overlaps the 170–180 dB range specified for small ships (lengths 55–85 m) by Richardson et al. (1995). When frequency dependent absorption is included, mean broadband source levels increase by 0.5–1 dB (we have limited the upper frequency to 40,000 Hz). Our range of mean values is similar to recent estimates of broadband source levels for similar-sized modern vessels, but for some classes other estimates are 1–11 dB higher than our estimates. Figure 2 depicts broadband SL statistics for each class we studied and juxtaposes the results from other studies of modern ships for comparable classes. Some of these studies are discussed below, partially to assess our results and partially to consider some of the common ways in which methods may differ between studies of ship noise: sample sizes, bandwidths, averaging times, calibration procedures, background

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 14/35 030 subtractions, absorption or other frequency-dependent corrections, geometric spreading rates, and ship characterization (e.g., classification, criteria for isolation, speed, and range). Compared with mean broadband source levels (20–30,000 Hz, TL of −15, absorption assumed negligible) computed by Bassett et al. (2012) our means are 0–6 dB lower, depending on the class. The comparatively low values of our means cannot be explained by distinct methodology; their study used a narrower broadband bandwidth and a lower (modeled) transmission loss. The most likely explanation for the differences in most classes is a difference in distinct ship design and/or operating characteristics between Puget Sound and Haro Strait populations. There is some evidence that ships measured by Bassett et al. (2012) may have higher speeds than in our study. Of the 24 select ships for which Bassett et al. (2012) provide speed data, 38% have SOG greater than 1 standard deviation above our mean values for the corresponding class. The average elevation of SOG for those ships is +2.0 m/s (+3.8 knots). Compared with broadband source levels (20–1,000 Hz, TL of −20) listed for 29 individual ships by McKenna et al. (2012) the mean values for equivalent classes in Table are 1–13 dB lower. These differences are also depicted in Fig. 2. Accounting for the difference in TL (1.4 dB/decade of range) between the studies would raise our SL values an average of 4.7 dB, thereby causing our inter-quartile range to overlap with or encompass the ranges of McKenna et al. (2012) for all comparable classes except bulk carriers. As with the Bassett et al. (2012) study, adjusting for differences in broadband bandwidth would raise their individual ship source levels even higher above our means, so cannot help explain the differences. Examining the SOG differences by class offers less of an explanation in this case; of the 29 ships, only 3 (about 10%) have speeds that exceed our mean SOG in the associated class, and only by an average of 1 m/s (2 knots). A study of 593 container ship transits by McKenna, Wiggins & Hildebrand (2013) yielded a mean source level (20–1,000 Hz, TL of −20) for the population of 185 dB, 7 dB higher than our mean of 178 dB for 529 container ship transits. In Supplemental Information 1, McKenna, Wiggins & Hildebrand (2013) provide a mean speed of 10.5 ± 1.4 m/s— roughly 0.5 m/s above our container ship mean speed of 9.9 m/s—and an mean range of 3,246 ± 291 m (about 1 km larger than our mean range). The speed difference could only account for about 0.5 dB of the source level discrepancy between the studies, based on the +2.2 dB per m/s (+1.1 dB/knot) relationship between broadband source level and speed portrayed for a single ship in McKenna, Wiggins & Hildebrand (2013). Compared with broadband source levels (45–7,070 Hz) of individual vessels measured by Malme, Miles & McElroy (1982) and Malme et al. (1989) and tabulated by Richardson et al. (1995) our means for respective classes are 1 dB lower than a tug (171 dB at 5.0 m/s (9.7 knots)), 6 dB lower than a cargo ship (181 dB), and 12 dB lower than a large tanker (186 dB). These differences might be due to more modern ships decreasing their speed (at least while in coastal waters) or increasing their propulsion efficiency. Kipple (2002) measured 6 cruise ships at a range of 500 yards and reported broadband source levels (10–40,000 Hz, TL of −20, absorption ignored) of 175–185 dB re 1 µPa 1 yard at 10 knots (5 m/s) and 178–195 dB re 1 µPa @ 1 yard at 14–19 knots (7–10 m/s). In comparison, our population of passenger ships (including cruise ships) has a mean SL

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 15/35 031 of 166 ± 8 dB re 1 µPa @ 1 m and a mean speed of 14.4 ± 4.5 knots. Thus, while the speeds tested by Kipple (2002) bracket our mean speed, our mean SL is 9–29 dB lower than their range of source levels. One possible explanation for this difference is an unspecified upward correction of received levels below 300 Hz that Kipple (2002) made to account for multipath propagation effects. This is substantiated by Malme et al. (1989), who state that passenger vessels in Southeast Alaska have SL from 170 to180 dB, a range that falls between our mean and maximum SL for passenger vessels and mostly below the ranges given by Kipple (2002). Finally, Arveson & Vendittis (2000) measured a bulk carrier at 8–16 knots (4–8 m/s) and found broadband source levels (3–40,000 Hz, TL of −20) of 178–192 dB. The source levels they calculated for speeds of 12 knots (6 m/s) and 14 knots (7 m/s), 184 dB and 190 dB, respectively, are most comparable to our bulk carrier population with SOG of 7.0 ± 0.7 m/s. Without correction for the different transmission loss assumptions, our bulk carrier SL of 173 ± 5 dB is 11–17 dB below their levels. While this pattern could be interpreted as an underestimation of SL by our methods, we believe our population statistics represent an accurate estimate of source levels for modern ships operating in coastal waterways. In almost all of the cases that we have discussed, the maximum discrepancy is less than 1.5 times the inter-quartile distance (25% vs 75% quantiles) for the comparable ship class (see Fig. 2). Exceptions are some of the louder container ships in McKenna, Wiggins & Hildebrand (2013) and vehicle carriers in McKenna et al. (2012), the large tanker mentioned in Richardson et al. (1995), the higher-speed cruise ships of Kipple (2002), and the bulk carrier of Arveson & Vendittis (2000) when its speed was greater than 8 knots (4 m/s). Even these exceptional upper values from the literature are almost completely contained within the distribution of our broadband SL population. Our maximum SL for a bulk carrier (191 dB) is 3.6 dB higher than the loudest bulk carrier tabulated in McKenna et al. (2012) and above the bulk carrier source levels obtained by Arveson & Vendittis (2000) at all speeds except 16 knots, or 8.2 m/s (192 dB). The loudest bulk carrier tabulated in Bassett et al. (2012) with source level of 182 dB is equal to the 95% quantile of SL within our bulk carrier class. The loudest ship tabulated by Richardson et al. (1995), a tanker with SL of 186 dB, is only 0.8 dB above our loudest tanker. One explanation for this outlier is that the ship was a supertanker driven by a steam-turbine—and therefore may represent the ‘‘upper range of large merchant vessels’’ (Malme et al., 1989). Finally, our passenger vessel population has a 95% quantile of 177 dB and a maximum of 183 dB, a range that encompasses most of the slow ships and the lower portion of the faster ships assessed by Kipple (2002). Across all classes, the maximum broadband SL for an individual ship was 195 dB for a container ship, 7 dB above the highest overall values reported by McKenna et al. (2012) and Bassett et al. (2012)— both for container ships, as well. Our maximum is consistent with the study of 593 container ships by McKenna, Wiggins & Hildebrand (2013) in which the maximum source level was also 195 dB. Our second- and third-highest maxima within a class were from a bulk carrier (191 dB) and a cargo ship (186 dB). All other classes had

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 16/35 032 maximum SL ≤ 185 dB m. The lowest maximum SL within a class was 176 dB for pleasure craft. The range of minimum broadband SL across all classes in our study was from 130 dB for a cargo ship to 167 dB for a vehicle carrier. In comparison, McKenna et al. (2012) reported a minimum SL across all classes of 177 dB for a chemical tanker, while the minimum SL for a container ship in McKenna, Wiggins & Hildebrand (2013) was 176 dB. In contrast with the exact agreement of the maxima between our container ship population and the data set of McKenna, Wiggins & Hildebrand (2013), this discrepancy of at least 9–10 dB in SL minima suggests that methodological differences between the studies may exert greater bias when ship signal levels are near background noise levels. Ship speed Averaged across all vessels, the SOG of isolated ships northbound in Haro Strait is 7.3 ± 2.0 m/s (14.4 ± 3.9 knots). This is higher than the mean of 10–12 knots (5.1–6.2 m/s) observed during WWII, but possibly lower than the post-war (mid-1970s) mean of about 15 knots (7.7 m/s) (Ross, 1976). In our study, the fastest classes are container ships (mean SOG of 9.9 m/s) and vehicle carriers (8.7 m/s), while the slowest vessels are fishing boats (4.7 m/s) and tugs (4.2 m/s). For tankers, our SOG of 7.1 ± 0.7 m/s is slightly below the 7.2–8.2 m/s (14–16 knot) range reported by Ross (1976) for both ‘‘T2 tankers’’ in WWII and supertankers built after about 1960. Overall, our data set samples a small range of ship speeds within any given class. Because Haro Strait is relatively long and straight, most vessels transit it without changing speed. Whether north- or south-bound, they have consistent SOG means and standard deviations. This low variability in speed limits our ability to search for relationships between noise and speed, but may help us discern in future work the influence of other variables—like propeller type, draft (loading), or maintenance levels—building on insights from McKenna, Wiggins & Hildebrand (2013).

Relationship between speed and broadband source level Upon linear regression of SL versus SOG for all data, we find a slope of +1.8 dB per m/s (+0.93 dB/knot). The coefficient of determination (R2) for this fit explains only 27% of the variance in the data (assuming normal distribution). Furthermore, most of the variation in SL is likely driven by ship class (which was not controlled for in the regression), with little change in speed within ship class. Slopes vary from +0.2 to +3.4 dB per m/s between ship classes. Examination of repeated transits of individual ships shows that the variation in slope is high between individual ships within a class and the percent of variance explained is low. While slopes are positive for most individual ships, some are zero or negative. These variations indicate that the overall population slope should not necessarily be applied to all ship classes or individual ships, echoing the recommendations of McKenna et al. (2012). Received spectra Most ships transiting Haro Strait raise background noise levels in the core summertime habitat of SRKWs at all measured frequencies (Fig. 3). Specifically, 95% of the ships

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 17/35 033 generate received spectrum levels at or above the 95% quantile of background levels from 20–96,000 Hz. Thus, at ranges of a couple kilometers, commercial ships cause significant underwater noise pollution not only at low frequencies, but also at high-frequencies. The difference in median spectrum levels between ship and background noise levels is more than 30 dB below 100 Hz and gradually decreases to about 10 dB at 20,000 Hz. In the high frequency range of 20,000–96,000 Hz the median ship noise is elevated above median background spectrum levels by at least 5 dB. This significant elevation of background levels at high frequencies is what motivated us to account for absorption when computing ship source levels and is consistent with an observation by Hildebrand et al. (2006) of a single commercial ship in Haro Strait at a range of 442 m that elevated the ambient noise spectrum levels by as much as 30–40 dB across a broad band of the spectrum (60–75,000 Hz). If we define the 5% quantile of background noise as an ‘‘ancient’’ ambient condition (Clark et al., 2009) then the typical (median) modern ship raises spectrum noise levels above ancient levels by 12–17 dB at frequencies used in killer whale echolocation (20,000–70,000 Hz) and by 17–35 dB at frequencies used in killer whale social vocalization (200–20,000 Hz). In the frequency range used by vocalizing baleen whales (20–200 Hz), the median ship spectrum noise levels are about 32–35 dB above the ancient ambient levels. We gain additional confidence in the accuracy of our sound pressure levels (and implicitly our system calibration) by comparing the received spectrum levels in Fig. 3 with ambient noise spectra from other studies. Our background quantiles are bracketed by the average deep-water ambient noise levels associated with sea state 1–3, though the slope of our median curve from 1,000–10,000 Hz is −8 dB/decade, about half as steep as the open-ocean slope of −17 dB/decade Urick (1983). The ‘‘usual lowest ocean noise’’ curve of Cato depicted in Plate 5 of National Research Council et al., (2003) is bounded by our 5 % and 25% quantiles from about 30 to 10,000 Hz. Two ambient noise spectra obtained in Haro Strait by Hildebrand et al. (2006) have levels that are bounded by our 5% and 95 % quantiles of background noise from 300 Hz to 30,000 Hz. The single ship spectrum (60 Hz–75,000 Hz) obtained opportunistically by Hildebrand et al. (2006) at a range of 442 m has levels that are greater than our 75% quantile of RLB at all frequencies. Similarly, our quantiles of total received spectrum level are consistent with previous studies. For example, the noise spectrum levels recorded in US bays and harbors during World War II by Urick (1983) are entirely bounded by our quantiles of RLT from 100 Hz to 10,000 Hz. The peak levels (at about 50 Hz) of the shipping contribution to deep water ambient noise estimated by Ross (1976) for ‘‘remote, light, moderate, and heavy’’ traffic are approximately 71, 77, 85, and 95 dB, respectively; the upper three traffic levels are encompassed by our 5% and 95% quantiles, while the ‘‘remote’’ levels are no more than 2 dB below our 5% quantile. Finally, the quantiles of unweighted received spectrum levels in Bassett et al. (2012) peak near 50 Hz and have levels that are within about 5 dB of our levels for corresponding quantiles at all frequencies common to the two studies. Even at high-frequencies our data are consistent; Knudsen, Alford & Emling (1948) reported total received levels of 40–50 dB at 30,000 Hz in coastal waters, a range which brackets our quantiles at that frequency.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 18/35 034 Source spectra Median source spectra for the whole ship population are shown in Fig. 4 as spectrum, 1/12-octave, and 1/3-octave levels, with and without accounting for absorption. For the spectrum levels, we also present 25% and 75% quantiles.

Source spectrum levels without absorption The median spectrum levels peak near 50 Hz at about 154 dB and decrease at higher frequencies with a slope of about −15 dB per decade (from 50–40,000 Hz). The 25% and 75% quantiles are 3–5 dB from the median below about 10,000 Hz, but at higher frequencies the difference decreases to about 1 dB re. In the region between 700 and 40,000 Hz the median spectrum has a subtle slope break near 5,000 Hz, with a slope of about −10 below and about −20 above. Previous observations, models, and experimental results all help contextualize these whole-population spectrum levels. Unfortunately, many previous studies of ship noise are not comparable due to presenting species-specific band levels (e.g., Hatch et al., 2012) or band levels rather than spectrum levels, or other limitations: small sample size, non- overlapping frequency ranges, and ship classes with low diversity, distinct definitions, or incomparable ships (e.g., ice breakers in Erbe & Farmer, 2000). One exception that allows comparison up to 1,200 Hz is the analysis of 54 ships at ranges of 360–1,800 m by Wales & Heitmeyer (2002). Their measured average source spectrum levels are bounded by our 25% and 75% quantiles from 400 to 1,200 Hz. At lower frequencies (below 400 Hz) their mean levels exceed our 75% quantile by 2–20 dB (20 dB at 20 Hz; 5 dB at 50 Hz; and 2 dB at 100 Hz). Interestingly, their curve does not peak near 50 Hz, but instead continues rising as the frequency decreases to 30 Hz, the lowest frequency they measured. The slope of their mean curve is about −30 dB/decade below 100 Hz, and −20 dB/decade above. They note that the variance around their mean levels decreases with rising frequency from a standard deviation as high as 5.32 dB below 400 Hz to about 3.12 dB above it. This suggests that a partial explanation for the elevation of their mean values relative to our 75% quantile may be variability in low-frequency power between ships. Models of ship noise that output spectrum levels provide another point of comparison. Our 50% and 75% quantiles are encompassed in the spectrum levels presented by National Research Council et al., (2003) for 3 classes of tankers, as well as merchant and fishing classes, based on the RANDI model (Wagstaff, 1973; Breeding et al., 1994) parameterized with data from Emery, Bradley & Hall (2001) and Mazzuca (2001). The 25% quantile is also encompassed, except below 30 Hz. Below 300 Hz, our median values lie between the fishing and merchant class levels of National Research Council et al., (2003); at higher frequencies—up to 1,000 Hz, the upper limit of their estimates—our median values are above their merchant class but below their intermediate tanker class (length 153–214 m, speed 7.7–9.3 m/s). Overall, this comparison suggests that our median source level spectra validate the RANDI model as parameterized in National Research Council et al., (2003) at intermediate frequencies (100–1,000 Hz), but below 100 Hz our median levels are lower

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 19/35 035 (by about 5–30 dB) than the RANDI model predicts for all classes except fishing vessels (length and speed bins of 15–46 m, 3.6–5.1 m/s). Other noticeable differences between our population median spectrum levels and those modeled in National Research Council et al., (2003) are the frequency of the peak power, the general slope of the spectra above the peak, and secondary peaks resolved in our data. While our spectra peak near 50 Hz, the peak power in the spectra of National Research Council et al., (2003) occurs slightly lower, at 30 Hz. Between 100 and 1,000 Hz, the slope of our median spectrum is −12 dB per decade, nearly three times less steep than the slope of −35 dB per decade in National Research Council et al., (2003). Our spectrum levels have detailed structure where the RANDI model curves of National Research Council et al., (2003) are smooth. Our quantiles show secondary power peaks between 80 and 1,100 Hz and many narrowband peaks in 1,100–10,000 Hz range, similar to the frequency dependence of spectral line complexity observed by Wales & Heitmeyer (2002). Experiments with cavitation provide a final comparison with our whole-population spectrum levels. Above 5,000 Hz the slope of our median spectrum matches the slope observed during cavitation of a spinning rod (Mellen, 1954) and a water jet (Jorgensen, 1961)—−20 dB per decade, (or −6 dB per octave).

Source spectrum levels with absorption The spectrum levels with absorption are indistinguishable from those without absorption below about 5,000 Hz. At higher frequencies, the SLa median spectrum level curve diverges from the SL curve, and starts to rise rapidly at the 40,000 Hz cut-off of this study. The associated 25% and 75% quantiles are within 3–5 dB of the median values throughout the region of divergence. These alternative source spectra look unfamiliar at high frequencies, and are not consistent with available data taken close (less than 500 m) to ships. For example, the single container ship measured at a range of 442 m by Hildebrand et al. (2006) in Haro Strait has a absorption-corrected source spectrum level of 108 dB re 1 µPa2/Hz @1 m at 40,000 Hz—about 17 dB below our SLa spectrum level at that frequency. However, we believe the absorption-corrected spectra in Fig. 4 are rooted in accurate physics and we note that the spectrum levels of SLa are in agreement with some measurements of underwater noise radiated during fully developed cavitation. For example Lesunovskii & Khokha (1968), specify rotating bar noise spectrum levels of 95–115 dB at 10,000 Hz while our 25%–75% quantiles of SLa spectrum level at that frequency are 114–120 dB. Similarly, Blake, Wolpert & Geib (1977) report noise levels from a cavitating hydrofoil of 75–110 dB re 1 µPa2/Hz @ 1 yd at 31,500 Hz which is approaching 2 our 25%–75% quantiles of SLa at that frequency (120–125 dB re 1 µPa /Hz @ 1 m). We expect that propeller cavitation noise intensity will be greater than laboratory measurements due to increased length scale and number of the blades on ships. Evidence from World War II studies of torpedo and submarine noise attributed to cavitation supports this expectation. Figures 10.21–10.23 of Urick (1983) show levels equivalent to 2 or bracketing our SLa spectrum levels: 24,000 Hz spectrum levels of 118 dB re 1 µPa / Hz @ 1 yd for a submarine cruising at 8 knots (4 m/s) near periscope depth; 25,000 Hz

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 20/35 036 spectrum levels of 100–130 dB re 1 µPa2/Hz @ 1 yd for torpedos moving at 20–45 knots (10–23 m/s); and 20,000 Hz spectrum levels of 115–130 dB re 1 µPa2/Hz @ 1 yd for a suite of torpedoes.

Source 1/12- and 1/3-octave levels The median 1/12- and 1/3-octave level curves in Fig. 4 are elevated relative to the median spectrum levels and diverge from them above 50 Hz due to the integration of spectrum levels over bands that get progressively wider with increasing center frequency. Like the spectrum levels, these curves have a peak near 50 Hz. Peak values are 158 dB re 1 µPa2 per band @ 1 m for the 1/12-octave levels and 163 dB for the 1/3-octave levels. Above 50 Hz, both curves have slopes of about −4 dB/decade from 100 to 5,000 Hz, −10 dB/decade from 5,000 to 40,000 Hz. While we are unaware of a comparable aggregation of source spectra from multiple ship classes presented as 1/3-octave levels, there are many studies of individual ships or classes that present 1/3-octave source levels. We compare them here with the median 1/3-octave curve for our ship population because we present only spectrum levels when assessing inter- and intra-class differences in subsequent sections. Our median 1/3-octave levels are entirely bounded by the estimated levels for 6 diverse ship types presented in Figure 3.14 of Malme et al. (1989) at all comparable frequencies (20–16,000 Hz). Similarly, our levels are within the estimated 1/3-octave source levels (10–10,000 Hz) summarized in Figure 6.5 of Richardson et al. (1995) for an ice breaker, a composite of supertankers, and a tug/barge at almost all frequencies. Only above about 2,000 Hz is our median curve slightly below comparable vessels described by Richardson et al. (1995): ours is within 2 dB of their tug/barge levels, and no more than 10 dB below their supertanker levels. Overall, we find the consistency of our results with these two studies to be remarkable. Comparing our median curve with the 7 ships (representing five of our classes) for which McKenna et al. (2012) presented 1/3-octave levels, our levels are 5–10 dB lower at all common frequencies (20–1,000 Hz). As discussed when presenting spectrum levels, we are not sure how to account for this difference, other than to recognize key differences between the studies: distinct transmission loss, our much larger sample size, and our higher diversity of classes. Studies of ship noise in which speed was varied present a range of levels that is also consistent with our results. Compared with the maximum–minimum envelopes of 1/3- octave source levels (referenced to 1 yard) from 6 cruise ships presented by Kipple (2002) our 1/3-octave levels are within the envelope for both 10 knot (5 m/s) and 14–19 knot (7.2–9.8 m/s) samples, except below 25 Hz where our levels are lower by 1–7 dB. Our levels also fall within (but near the lower edge) of the range of 1/3-octave spectra reported by Arveson & Vendittis (2000) for a bulk carrier tested from 68 to 148 rpm. Our 1/3-octave levels help validate the RANDI model used by Erbe, MacGillivray & Williams (2012) to compute 1/3-octave spectra for five ship length classes over a range of speeds observed in traffic off the coasts of British Columbia and Washington State. Overall, our median levels are entirely within the range of their estimated levels at all

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 21/35 037 modeled frequencies (10–2,000 Hz). More specifically, though, our median crosses their size-specific curves, because it has a less steep slope. Below 400 Hz our levels are bounded by their L1 and L3 classes (representing lengths less than 50 m); above 400 Hz our median levels are between their L4 and L5 classes (greater than 50 m). An even more dramatic crossing of model curves by our median 1/3 octave source spectrum level curve is evident upon comparison with Figure 1 of Williams et al. (2014). While our median source levels are equivalent to or bounded by the 1/3-octave levels for each of their modeled ship types (tug, cruise ship, container ship) near or below 250 Hz, at higher frequencies our levels exceed the modeled ones by 7–10 dB. The crossing of such modeled spectra by our 1/3-octave median curve is one manifestation of a subtle slope difference between our results and previous studies (Arveson & Vendittis, 2000; Kipple, 2002; Erbe, MacGillivray & Williams, 2012; Williams et al., 2014). While Arveson & Vendittis (2000) observe slopes from above a 55 Hz cavitation ‘‘hump’’ up to about 30,000 Hz to be −10 dB/decade on a 1/3-octave plot, our slope over the same frequency range is shallower (−6.5 dB/decade) and we observe a slope break near 3,000 Hz. Below the break the slope is about −4.5 dB/decade, while above it is −10 dB/decade. The similarity of our 1/3-octave levels with those from available studies at frequencies below 630 Hz (the lowest tone used in our transmission loss experiment) is the first evidence that our measurements of low-frequency radiated noise are accurate. The lower slope relative to other studies suggests that the ship population in this study is generating proportionally more high-frequency noise than ships in previous studies.

Source 1/12- and 1/3-octave levels with absorption As with the spectrum levels, the 1/12- and 1/3-octave level curves with absorption are indistinguishable from those without absorption below 5,000 Hz. At higher frequencies, the SLa median 1/12- and 1/3-octave levels rise to match the 50 Hz levels of the associated median SL curves near 35,000 Hz and then continue to increase at higher frequencies. This means that when we account for absorption when computing 1/12- or 1/3-octave levels, modern ships radiate noise in high-frequency bands (centered near 35,000 Hz) at levels equivalent to the low-frequency maxima near 50 Hz. This surprising equivalency, and the theoretically even higher power levels in bands above 35,000 Hz, are important to consider when assessing the masking potential of ship noise in habitats close to or within shipping lanes for marine species that utilize high-frequency signals. Although it is novel to state that ship noise source levels have peak power at high- as well as low-frequencies, we provide these 1/12- and 1/3-octave noise levels to facilitate accurate modeling of acoustic impacts for species that have critical bands overlapping these octave bands (Richardson et al., 1995). While the median 1/12-octave source levels reported by Erbe & Farmer (2000) for the cavitating propeller of an ice breaker are not comparable to any of our ship classes (and much higher—30 dB re 1 µPa2 per band @ 1 m above our median level at their power peak near 500 Hz), we note that the slope of their median curve is −13 dB/decade from 1,000 to 10,000 Hz. Importantly, Erbe & Farmer (2000) is rare in stipulating that absorption

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 22/35 038 Figure 5 Median source spectra of ship noise for different classes of ships. Comparison of median source spectrum levels (without absorption) between ship classes.

was accounted for in computing source levels. Their slope is about twice as steep as our 1/12-octave median slope of −7 dB/decade in the same frequency range. Finally, Kipple (2002) did not correct for absorption, but made measurements of cruise ship receive levels up to 40,000 Hz at short range (500 yards) and therefore provides a rare point of reference for our high-frequency SLa levels. Their 1/3-octave band source levels at 40,000 Hz for a suite of cruise ships and speeds (10–19 knots; 5.1–9.7 m/s) vary from 133 to 154 dB re 1µPa @ 1 yard—values that approximately bracket our uncorrected SL levels but are 13–34 dB below our SLa levels. Spectral diﬀerences between ship classes When the ship population is broken down by class (Fig. 5) the medians show a striking bifurcation. While all classes have similar median spectrum levels near 20,000 Hz, the curves diverge at lower frequencies, and below 200 Hz they bifurcate into high- and low-power groups. The high-power group has peak power of 153–159 dB near 50 Hz (just above the population median shown in Fig. 4) and consists of container ships, vehicle carriers, cargo ships, bulk carriers, and tankers. The low-power group has peak power of 134–141 dB near 50 Hz or just above 100 Hz—levels well below the population median or even 25% quantile—and consists of passenger vessels, tugs, military, research, fishing, miscellaneous, and pleasure vessels. The 25%, median, and 75% spectrum levels at the power peak near 50 Hz in Fig. 4 bracket the 50 Hz levels of the high-power group of ships in Fig. 5. The median of the whole population is most similar to the spectra in the high-power group (e.g., the bulk carrier curve) because the aggregated sample size is much higher in the high-power group

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 23/35 039 than in the low-power one (see Table 2). Modelers interested in assessing impacts of specific ship classes, particularly those in the lower-power group, should not use the median or 25% quantile levels for the whole population, but instead select class-specific levels from the curves in Fig. 5. Container ships have the highest median source spectrum level of all classes at almost all frequencies below 10,000 Hz with peak power of 159 dB near 40 Hz. This is likely because of their relatively large size and high mean speed (10 m/s) compared to pleasure craft or military ships—the classes with the lowest median power at all frequencies below 400 Hz. Many of the ship classes show secondary peaks in the median spectrum levels from 100 to 5,000 Hz. For example, most classes show a 2 dB dip near 250 Hz and at least container ships, vehicle carriers, cargo ships, and tankers have peaks near 300, 700, and 1,000 Hz. There are also narrower peaks for these same classes between 1,000 and 10,000 Hz, most prominently at 2,000 Hz and near 3,000 Hz. The variability of the median source level in each class decreases above 5,000 Hz and remains low until about 10,000 Hz. At higher frequencies (10,000–40,000 Hz) the variability increases again for most ship classes, but the degree of increase is a strong function of sample size within a class. While we know from examining spectrograms from individual ships that some of the narrow peaks are associated with active acoustic sources (depth sounders, scientific echosounders, and fish finders), in Fig. 5 the high variance above 10,000 Hz is due primarily to some ships having spectrum levels that do not meet the robust threshold at higher frequencies. Particularly in classes where the sample size is already small this leads to some high frequency bins having many fewer data points than adjacent bins which in turn results in more-variable median values across this high-frequency range. The quantiles of source spectrum level by class in Fig. 6 provide further detail about inter-class differences. Comparing the 95% quantiles, container ships still have the highest peak power (165 dB) near 50 Hz, but bulk and vehicle carriers, cargo ships and tankers also have peak power greater than 160 dB. Other classes have peak power in the 95% quantiles near 50 Hz at spectrum levels that range from 156 dB (research) to 150 dB (tugs). Comparing the 5% quantiles, we expected that the military class would have the lowest levels due to more advanced ship-quieting technologies. While the military class levels are much lower than container ships (10 dB less at 1,000 Hz and 20 dB less at 100 Hz), other classes have even lower levels at those frequencies, particularly fishing vessels and pleasure craft.

Spectral variability within ship classes All classes of ships have spectrum levels that vary more at low frequencies than at high frequencies (Fig. 6). Near 50 Hz there is a 15–35 dB difference between the 5% and 95% quantile levels. That difference decreases with rising frequency until above 20,000 Hz it is typically less than 10 dB. Below 20,000 Hz, source level variability in Fig. 6 tends to be lower for the classes that have smaller speed over ground standard deviations and that have larger sample size as shown in Table 2. While container and cargo ships, bulk and vehicle carriers, and tankers have 95–5% spectrum level differences of about 15 dB, the other classes exhibit larger

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 24/35 040 Figure 6 Quantiles of ship source spectra for different classes of ships. Quantiles of source spectrum levels for each class of ship. Median (50%) quantile (black) overlies 5, 25, 75, and 95% quantiles (blue).

differences up to 25–30 dB. The classes with the largest number of vessels are most uniform in their speed over ground and most consistent in their vessel design and operation. Tugs are a special case because there are many transits and their speed is not unusually variable, but their loading is. Our passenger vessels are all cruise ships and hence similar in design, but their speeds are quite variable as they adjust their arrival times in the Port of Vancouver. Finally, the small numbers of pleasure craft and vessels classed as miscellaneous are highly variable in both their designs and their operations. Other studies have observed a similar pattern of source level variability with frequency. In mean source spectrum levels from 54 ships Wales & Heitmeyer (2002) noted higher, more-variable standard deviations from 30 to 400 Hz and lower, more-constant ones from 400 to 1,200 Hz. Figure 8 of McKenna, Wiggins & Hildebrand (2013) displays histograms of octave-band power for 593 container ships which have widths that decrease from about 35 dB in the 16 Hz band to 26 dB in the 500 Hz band. One explanation for this pattern is that the low-frequency portion of ship noise spectra is influenced by diverse design and operational details (many sources of variability), while cavitation generates high-frequency broadband noise (including up to 100,000 Hz) no

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 25/35 041 matter its source. As mentioned in the introduction, there are many sources of ship noise below 1,000 Hz that should be expected to vary between individual ships in a particular class. Conversely, a wide range of vessels have been documented to radiate elevated high- frequency noise upon increased engine RPM or SOG—conditions reasonably associated with increased cavitation (Erbe & Farmer, 2000; Kipple, 2002; Hildebrand et al., 2006). The literature offers a handful of spectra for particular classes that can be compared with the quantiles of Fig. 6. These spectra typically come from individual ships, though, so can only serve to verify the range of our quantiles, rather than assessing the accuracy of the quantiles themselves. The spectrum levels provided by McKenna et al. (2012) for individual ships in comparable classes (a container ship, a vehicle carrier, two bulk carriers, and a few tankers) all fall within a few dB of our 95% quantile. Only their bulk carrier deviates from this pattern with levels near 100 Hz higher by about 10 dB. Overall, the broadband and spectrum levels of ships associated with the port of Los Angeles (McKenna et al., 2012) are most comparable to the noisiest 5% of ships transiting Haro Strait. Similarly, the source spectrum levels for a single container ship measured in the middle of Haro Strait by Hildebrand et al. (2006) also fall within the 5% and 95% quantiles of our cargo class (from 90 Hz to 40,000 Hz). The alignment of such individual ship spectra within the quantiles of their associated class at all common frequencies—and most importantly at frequencies below that of our lowest transmission loss test tone—helps verify our extrapolation of the near-spherical spreading we observed from 630 to 20,100 Hz to all frequencies reported in our study. We take this spectral consistency across multiple classes as evidence that the ship noise received at our nearshore hydrophone has not undergone shallow water attenuation. While normal mode theory (Urick, 1983) would predict a cutoff frequency of about 50 Hz if our hydrophone were in a shallow channel 8 m deep, that is not the bathymetric situation at our study site. Instead, Haro Strait is a 250–300 m deep channel with a steep western wall of sparsely sedimented solid rock (Jones & Wolfson, 2006) and our hydrophone is positioned near the top of the wall where the offshore bottom slope is 20–30◦. In this situation, Jones & Wolfson (2006) expect not only destructive interference at ranges much greater than the source depth, but also upslope enhancement. In our transmission loss experiment, we did not observe any frequency dependent attenuation consistent with these phenomena. Furthermore, the theoretical cutoff frequency for a 250 m deep channel is 1.5 Hz (Urick, 1983), well below our lowest measured frequency band. We therefore argue that any effects of interference or backscatter are averaged out in our study, primarily because each isolated ship ensonifies the full width of this reverberating channel and moves 150–300 m during a 30-second recording (1–2 times the 130 m wavelength or our lowest measured frequency, 11.5 Hz).

CONCLUSIONS Having ensured our samples were isolated (uncontaminated by noise from other ships or boats) and subtracted estimated background levels, we are confident that median

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 26/35 042 received levels of ship noise in the core of SRKW critical habitat are elevated above median background levels not only at low frequencies (20–30 dB from 100 to 1,000 Hz), but also at high frequencies (5–13 dB from 10,000 to 40,000 Hz). Thus, underwater noise radiated by modern ships extends to high frequencies just as boat noise does (Erbe, 2002; Kipple & Gabriele, 2004; Hildebrand et al., 2006). Earlier studies have also observed this aspect of ship noise, but with smaller sample size, over different frequency ranges and less diverse ship classes (Kipple & Gabriele, 2004; Hildebrand et al., 2006; Bassett et al., 2012), and/or in received rather than source levels (Hermannsen et al., 2014). Such ship noise has the potential to mask odontocete signals, especially in coastal environments where shipping lanes are close enough to the shoreline (<10 km) that high frequency sound is not fully absorbed. In the summertime habitat of the endangered SRKWs ship noise may interfere not only with SRKW communication (vocalizations) but also foraging and navigation (echolocation clicks). Average broadband received levels (11.5–40,000 Hz) for the entire ship population are 110 ± 7 dB and ranged from 97 ± 10 dB for pleasure craft to 116 ± 4 dB for container ships. The range of RL for container ships (112–120 dB) show that levels received by SRKWs along the coastline at Lime Kiln from some container ships occasionally meet or exceed the 120 dB broadband threshold currently used by NOAA to define level B harassment from non-impulsive noise in the US Ships northbound in Haro Strait exhibit moderate speeds with low variability (SOG of 7.3 ± 2.0 m/s or 14.1 ± 3.9 knots). Nevertheless, there is enough variation in speed across the whole population to reveal a linear relationship between received level and speed with a slope of +1.8 dB per m/s. This suggests a potential mitigation strategy for the average ship—slowing down—that has been recommended previously as an operational ship quieting option (Southall & Scholik-Schlomer, 2008). This strategy has other environmental benefits, like reducing collision risks, and is consistent with recent industry efforts to increase fuel efficiency (e.g., the ‘‘slow steaming’’ initiative of Maersk). For a passenger ship measured at speeds of 9–18 knots (4.6–9.3 m/s) during WWII Ross (1976) shows in Figure 8.19 that reducing speed lowers source spectrum levels by at about 1.5 dB/knot (2.9 dB per m/s) at all frequencies, but most notably lowers them by about 3.0 dB/knot (5.8 dB per m/s)—both at high frequencies (above 10,000 Hz) and at low frequencies (less than 100 Hz). Average broadband source levels were 173 ± 7 dB for the population. Comparing broadband source levels between ship classes, container ships have the highest mean SL of 178 ± 4 dB. Therefore, assuming near-spherical transmission loss, marine life within a couple kilometers of shipping lanes will commonly receive noise levels above NOAA’s 120 dB threshold. At ranges less than about a kilometer, receive levels from many ships in Haro Strait will exceed the 130–150 dB modeled ship noise (10–50,000 Hz) dose associated with minor changes in northern resident killer whale behavior (Williams et al., 2014). At distances of less than about a kilometer, it is likely that received 1/12- or 1/3-octave band levels at high frequencies are equal or greater than they are at low frequencies. Further research should measure ship spectrum levels at ranges of a few hundred meters in order

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 27/35 043 to more fully quantify the high frequency (40,000–100,000 Hz) components of ship sound signatures. Models of noise impacts in habitat containing shipping lanes will be more accurate if parameterized with spectral data, as opposed to broadband levels. Since we observe spectral variability between and within the 12 classes of vessels in this study, most prominently the bifurcation at low frequencies between classes, such models should use the class-specific spectrum level quantiles if possible, rather than the whole-population spectrum and band level medians we have presented. Our broadband, spectrum, 1/12-octave, and 1/3-octave source levels for the whole population have median values that are comparable to the literature, with a few exceptions that we believe are due primarily to methodological differences. Some past analyses may not have made all recommended corrections (TC43 Acoustics, 2012); most commonly, methods sections are ambiguous about the definition and subtraction of background noise levels from total received levels prior to source level computations. It is also possible that these exceptions are due to sampling ship populations that are distinct (being composed of different individual ships/classes and/or operating differently). Even though our sample size is larger than most previous studies, we estimate that we sampled only about 1.6% of the 86,942 ships in the 2012/2013 global fleet (UNCTD, 2013). In any case, since our source level quantiles have slightly lower levels than some studies, particularly at low frequencies, they can be taken as a conservative characterization of the current fleet when developing ship noise models or policies. One subtle pattern we note is that compared to some previous measurements and models, our median source spectrum levels are relatively low below 200 Hz and relatively high above 20,000 Hz. One implication of this is that noise models using previous measurements may overestimate the low-frequency noise levels of some ship types and underestimate high-frequency noise levels. Such flattening of the spectral slope in more modern ships is described in Figure 8.20 of Ross (1976) which shows source spectrum levels (below 100 Hz and from 1,000 to 20,000 Hz) elevated 1–3 dB in large populations of post-War versus WWII-era vessels. Some studies show a flattening of spectra above 100–1,000 Hz as ship and engine speed increases (Ross, 1976; Arveson & Vendittis, 2000; Kipple, 2002). We speculate that this historical trend may be continuing and recommend further investigation of the evolution of both ship speed (Leaper, Renilson & Ryan, 2014) and the mitigation of low-frequency internal noise on ships for human health reasons. We recommend that future ship noise studies statistically characterize populations of ships—both their broadband and spectrum source levels. Having struggled to discern which studies in the literature are comparable to our results, we also suggest that future method sections be explicit about ship classification, calibration procedures, background subtraction and/or criteria for isolation from other sources, models and/or measurements of transmission loss, band width(s) and centers, absorption, and any other corrections. Metadata should include statistical representations of ship speeds and measurement ranges. Many studies are ambiguous about some of these details which complicates replication, comparison of results, and formation of hypotheses about observed differences.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 28/35 044 Future work should also assess covariates other than speed, such as size, as well as azimuthal and temporal variability in source spectrum levels. We know from years of listening to live audio streams of Salish Sea ship noise (free via orcasound.net) that there is great temporal variability in the noise radiated by many ships. A small percentage of ships emit periodic strong mid-frequency tones that are likely caused by singing propellers (Ross, 1976). Our next step is to explore such temporal variations in amplitude and frequency, identify statistical outliers that may represent extreme masking cases, and further investigate possible governing variables, including speed, class, azimuth, and loading. The variability we observe within ship classes indicates opportunities for reducing noise in ships, particularly those associated with the upper quantiles in each class. While the details of the spectral and temporal variability of noise from an individual ship may be important to a receiving species, metrics for measuring and regulating underwater noise will practically involve some temporal averaging, and possibly integration over bands wider than 1 Hz. We suggest a reasonable time scale for averaging ship noise is seconds or minutes, rather than a year as stipulated in the European Union’s Marine Strategy Framework Directive 2008/56/EC (Tasker et al., 2010). Additionally, based on the received signal above background noise that we observe at high frequencies, we recommend that future guidelines for monitoring ship noise raise the upper frequency limit of recording systems from 20,000 Hz (Dekeling et al., 2014) to at least 50,000 Hz. As Registered Ship Classification Societies continue to issue underwater radiated noise notations, we hope that these data can be used to assess their validity.

ACKNOWLEDGEMENTS We would like to thank all who helped deploy and maintain the calibrated hydrophone system. Logistical support was provided by The Whale Museum (Jenny Atkinson and Eric Eisenhardt), Beam Reach, intrepid divers (David Howitt), and SMRU Consulting. Analysis was accomplished through open-source software and data including: Generic Mapping Tools (GMT), NOAA bathymetry and shipping lanes, Qt, R, ggplot, Libreoffice, Overleaf, and Zotero. Chris Bassett and Marla Holt kindly provided helpful reviews of the pre-print; Michael Jasny and Hussein Alidina helped us understand the policy implications of our work; Leslie Veirs, Wendy Wood, and Annie Reese provided unflinching encouragement throughout. Finally, we thank the libraries of the University of Washington and Friday Harbor Labs for access to otherwise closed-access journals.

ADDITIONAL INFORMATION AND DECLARATIONS

Funding Funding for the Salish Sea hydrophone network came via Brad Hanson of the Northwest Fisheries Science Center and Lynne Barre of NOAA’s Western Regional Center, Washington State Parks, and Chuck Greene of Cornell University. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 29/35 045 Grant Disclosures The following grant information was disclosed by the authors: Northwest Fisheries Science Center. NOAA’s Western Regional Center. Washington State Parks. Chuck Greene of Cornell University. Competing Interests The authors declare there are no competing interests. Author Contributions • Scott Veirs and Val Veirs conceived and designed the experiments, performed the experiments, analyzed the data, contributed reagents/materials/analysis tools, wrote the paper, prepared figures and/or tables, and reviewed drafts of the paper. • Jason D. Wood conceived and designed the experiments, performed the experiments, analyzed the data, contributed reagents/materials/analysis tools, wrote the paper, and reviewed drafts of the paper. Data Availability The following information was supplied regarding data availability: http://www.beamreach.org/data/staff-research/ship-noise/. Supplemental Information Supplemental information for this article can be found online at http://dx.doi.org/10.7717/ peerj.1657#supplemental-information.

REFERENCES Aguilar Soto N, Johnson M, Madsen PT, Tyack PL, Bocconcelli A, Fabrizio Bor- sani J. 2006. Does intense ship noise disrupt foraging in deep-diving Cuvier’s beaked whales (Ziphius cavirostris)? Marine Mammal Science 22(3):690–699 DOI 10.1111/j.1748-7692.2006.00044.x. Andrew RK, Howe BM, Mercer JA, Dzieciuch MA. 2002. Ocean ambient sound: comparing the 1960s with the 1990s for a receiver off the California coast. Acoustics Research Letters Online 3(2):65–70 DOI 10.1121/1.1461915. Arveson PT, Vendittis DJ. 2000. Radiated noise characteristics of a modern cargo ship. The Journal of the Acoustical Society of America 107(1):118–129 DOI 10.1121/1.428344. Au WWL, Ford JKB, Horne JK, Allman KAN. 2004. Echolocation signals of free- ranging killer whales (Orcinus orca) and modeling of foraging for chinook salmon (Oncorhynchus tshawytscha). The Journal of the Acoustical Society of America 115(2):901–909 DOI 10.1121/1.1642628. Au WW, Pack AA, Lammers MO, Herman LM, Deakos MH, Andrews K. 2006. Acoustic properties of humpback whale songs. The Journal of the Acoustical Society of America 120(2):1103–1110 DOI 10.1121/1.2211547.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 30/35 046 Ayres KL, Booth RK, Hempelmann JA, Koski KL, Emmons CK, Baird RW, Balcomb- Bartok K, Hanson MB, Ford MJ, Wasser SK. 2012. Distinguishing the impacts of inadequate prey and vessel traffic on an endangered killer whale (Orcinus orca) population. PLoS ONE 7(6):e36842 DOI 10.1371/journal.pone.0036842. Bassett C, Polagye B, Holt M, Thomson J. 2012. A vessel noise budget for Admiralty Inlet, Puget Sound, Washington (USA). The Journal of the Acoustical Society of America 132(6):3706–3719 DOI 10.1121/1.4763548. Blake WK, Wolpert MJ, Geib FE. 1977. Cavitation noise and inception as influenced by boundary-layer development on a hydrofoil. Journal of Fluid Mechanics 80(04):617–640 DOI 10.1017/S0022112077002390. Breeding JE, Pflug LA, Bradley M, Hebert M, Wooten M. 1994. RANDI 3.1 user’s guide. Technical Report, DTIC document. Available at http://www.dtic.mil/dtic/tr/fulltext/ u2/a286300.pdf . Buckstaff KC. 2004. Effects of watercraft noise on the acoustic behavior of bottlenose dolphins, Tursiops truncatas, in Sarasota Bay, Florida. Marine Mammal Science 20(October):709–725 DOI 10.1111/j.1748-7692.2004.tb01189.x. Cerchio S, Jacobsen JK, Norris TF. 2001. Temporal and geographical variation in songs of humpback whales, Megaptera novaeangliae: synchronous change in Hawaiian and Mexican breeding assemblages. Animal Behaviour 62(2):313–329 DOI 10.1006/anbe.2001.1747. Clark CW, Ellison W, Southall B, Hatch L, Van Parijs S, Frankel A, Ponirakis D. 2009. Acoustic masking in marine ecosystems: intuitions, analysis, and implication. Marine Ecology Progress Series 395:201–222 DOI 10.3354/meps08402. Dekeling R, Tasker M, Graaf SVD, Ainslie M, Andersson M, André M, Borsani JF, Brensing K, Castellote M, Cronin D, Dalen J, Folegot J, Leaper R, Pajala J, Redman P, Robinson SP, Sigray P, Sutton G, Thomsen F, Werner S, Wittekind D, Young JV. 2014. Monitoring guidance for underwater noise in European seas-part II: monitoring guidance specifications. JRC Scientific and Policy Report UR 26555 EN, Publications Office of the European Union, Luxembourg. Dow MT, Emling JW, Knudsen VO. 1945. Survey of underwater sound No. 4: sounds from surface ships. Washington, D.C.: National Defense Research Committee, 6, Sec. 6.1, NDRC-2124. Emery L, Bradley M, Hall T. 2001. Data base description (DBD) for the historical temporal shipping data base (HITS), Version 4.0. In: Planning systems incorporated, Report TRS-301. Slidell: Planning Systems Inc. Erbe C. 2002. Underwater noise of whale-watching boats and potential effects on killer whales (Orcinus orca), based on an acoustic impact model. Marine Mammal Science 18(2):394–418 DOI 10.1111/j.1748-7692.2002.tb01045.x. Erbe C. 2011. Underwater acoustics: noise and the effects on marine mammals. Brisbane: JASCO Applied Sciences. Erbe C, Farmer DM. 2000. Zones of impact around icebreakers affecting beluga whales in the Beaufort Sea. The Journal of the Acoustical Society of America 108:1332–1340 DOI 10.1121/1.1288938.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 31/35 047 Erbe C, MacGillivray A, Williams R. 2012. Mapping cumulative noise from shipping to inform marine spatial planning. The Journal of the Acoustical Society of America 132(5):EL423–EL428 DOI 10.1121/1.4758779. Ford JKB. 1987. A catalogue of underwater calls produced by killer whales (Orcinus orca) in British Columbia. Vancouver: Department of Fisheries and Oceans, Canadian Data Report of Fisheries and Aquatic Sciences. Ford JKB, Ellis GM. 2006. Selective foraging by fish-eating killer whales. Marine Ecology Progress Series 316:185–199 DOI 10.3354/meps316185. Francois RE, Garrison GR. 1982. Sound absorption based on ocean measurements. Part II: boric acid contribution and equation for total absorption. The Journal of the Acoustical Society of America 72(6):1879–1890 DOI 10.1121/1.388673. Gray LM, Greeley DS. 1980. Source level model for propeller blade rate radiation for the world’s merchant fleet. The Journal of the Acoustical Society of America 67(2):516–522 DOI 10.1121/1.383916. Hall JD, Johnson CS. 1972. Auditory thresholds of a killer whale Orcinus orca Linnaeus. The Journal of the Acoustical Society of America 51(2B):515–517 DOI 10.1121/1.1912871. Hanson MB, Baird RW, Ford JKB, Hempelmann-Halos J, Van Doornik D, Candy J, Emmons CK, Gregory S, Brian G, Katherine A, Samuel W, Kelley BB, John S, Michael F, Ford. 2010. Species and stock identification of prey consumed by endangered southern resident killer whales in their summer range. Endangered Species Research 11:69–82 DOI 10.3354/esr00263. Hatch LT, Clark CW, Van Parijs SM, Frankel AS, Ponirakis DW. 2012. Quantifying loss of acoustic communication space for right whales in and around a US National Marine Sanctuary. Conservation Biology 26(6):983–994 DOI 10.1111/j.1523-1739.2012.01908.x. Hauser DDW, Logsdon MG, Holmes EE, VanBlaricom GR, Osborne RW. 2007. Summer distribution patterns of southern resident killer whales Orcinus orca: core areas and spatial segregation of social groups. Marine Ecology Progress Series 351:301–310 DOI 10.3354/meps07117. Hermannsen L, Beedholm K, Tougaard J, Madsen PT. 2014. High frequency components of ship noise in shallow water with a discussion of implications for harbor porpoises (Phocoena phocoena). The Journal of the Acoustical Society of America 136(4):1640–1653 DOI 10.1121/1.4893908. Hildebrand JA. 2009. Anthropogenic and natural sources of ambient noise in the ocean. Marine Ecology Progress Series 395:5–20 DOI 10.3354/meps08353. Hildebrand J, McDonald MA, Calambokidis J, Balcomb K. 2006. Whale watch vessel ambient noise in the Haro Strait. Technical Report MPL TM-490. Holt MM, Noren DP, Veirs V, Emmons CK, Veirs S. 2009. Speaking up: killer whales (Orcinus orca) increase their call amplitude in response to vessel noise. The Journal of the Acoustical Society of America 125:EL27–EL31 DOI 10.1121/1.3040028.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 32/35 048 Jones CD, Wolfson MA. 2006. Acoustic environment of Haro Strait: preliminary propagation modeling and data analysis. Technical report, DTIC document. Applied Physics Laboratory, University of Washington, Seattle. Jorgensen DW. 1961. Noise from cavitating submerged water jets. The Journal of the Acoustical Society of America 33(10):1334–1338 DOI 10.1121/1.1908433. Kipple B. 2002. Southeast Alaska cruise ship underwater acoustic noise. Technical Report NSWCCD-71-TR-2002/574, Naval Surface Warfare Center – Detach- ment Bremerton. Available at http://www.nps.gov/glba/naturescience/upload/ CruiseShipSoundSignaturesSEAFAC.pdf . Kipple B, Gabriele C. 2004. Underwater noise from skiffs to ships. In: Proceedings of the fourth Glacier Bay science symposium. Available at http://www.nps.gov/glba/ naturescience/loader.cfm?csModule=security/getfile&PageID=945733 . Knudsen VO, Alford RS, Emling JW. 1948. Underwater ambient noise. Journal of Marine Research 7:410–429. Leaper R, Renilson M, Ryan C. 2014. Reducing underwater noise from large commercial ships: current status and future directions. Journal of Ocean Technology 9(1):00004. Lesunovskii VP, Khokha YV. 1968. Characteristics of the noise spectrum of hydrodynamic cavitation on rotating bars in water. Akusticheskii Zhurnal 14:566–571. Malme CI, Miles PR, McElroy PT. 1982. The acoustic environment of humpback whales in Glacier Bay and Frederick Sound/Stephens Passage, Alaska. 4848, NMFS/MML. Cambridge: Bolt, Beranek, and Newman, Inc.. Malme CI, Miles PR, Miller GW, Richardson WJ, Roseneau DJ, Thomson DH, Greene Jr CR. 1989. Analysis and ranking of the acoustic disturbance potential of petroleum industry activities and other sources of noise in the environment of marine mammals. Technical Report 6945. BBN Systems and Technologies Corporation, Cambridge. Mazzuca LL. 2001. Potential effects of low frequency sound (LFS) from commercial vessels on large whales. PhD thesis, University of Washington. McDonald MA, Hildebrand JA, Wiggins SM. 2006. Increases in deep ocean ambient noise in the Northeast Pacific west of San Nicolas Island, California. The Journal of the Acoustical Society of America 120(2):711–718 DOI 10.1121/1.2216565. McKenna MF, Ross D, Wiggins SM, Hildebrand JA. 2012. Underwater radiated noise from modern commercial ships. The Journal of the Acoustical Society of America 131(1):92–103 DOI 10.1121/1.3664100. McKenna MF, Wiggins SM, Hildebrand JA. 2013. Relationship between container ship underwater noise levels and ship design, operational and oceanographic conditions. Scientific Reports 3: Article 1760 DOI 10.1038/srep01760. Melcon ML, Cummins AJ, Kerosky SM, Roche LK, Wiggins SM, Hildebrand JA. 2012. Blue whales respond to anthropogenic noise. PLoS ONE 7(2):. Mellen RH. 1954. Ultrasonic spectrum of cavitation noise in water. The Journal of the Acoustical Society of America 26(3):356–360 DOI 10.1121/1.1907341.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 33/35 049 Merchant ND, Pirotta E, Barton TR, Thompson PM. 2014. Monitoring ship noise to assess the impact of coastal developments on marine mammals. Marine Pollution Bulletin 78(1–2):85–95 DOI 10.1016/j.marpolbul.2013.10.058. Mooney TA, Yamato M, Branstetter BK. 2012. Hearing in cetaceans: from natural history to experimental biology. In: Advances in marine biology, vol. 63. New York: Academic Press, 197–246. Morisaka T, Shinohara M, Nakahara F, Akamatsu T. 2005. Effects of ambient noise on the whistles of Indo-Pacific bottlenose dolphin populations. Journal of Mammalogy 86(3):541–546 DOI 10.1644/1545-1542(2005)86[541:EOANOT]2.0.CO;2. Munger LM, Wiggins SM, Moore SE, Hildebrand JA. 2008. North Pacific right whale (Eubalaena japonica) seasonal and diel calling patterns from long-term acoustic recordings in the southeastern Bering Sea, 2000–2006. Marine Mammal Science 24(4):795–814. National Research Council, Committee on potential impacts of ambient noise in the ocean on marine mammals, ocean studies board, and division on earth and life studies. 2003. Ocean noise and marine mammals. Washington, D.C.: National Academies Press. Richardson WJ, Greene Jr CR, Malme CI, Thomson DH. 1995. Marine mammals and noise. New York: Academic Press. Riesch R, Ford J, Thomsen F. 2006. Stability and group specificity of stereotyped whistles in resident killer whales, Orcinus orca, off British Columbia. Animal Behaviour 71(1):79–91 DOI 10.1016/j.anbehav.2005.03.026. Rolland RM, Parks SE, Hunt KE, Castellote M, Corkeron PJ, Nowacek DP, Wasser SK, Kraus SD. 2012. Evidence that ship noise increases stress in right whales. Proceedings of the Royal Society B: Biological Sciences. Ross D. 1976. Mechanics of underwater noise. Oxford: Pergamon Press. Southall BL, Bowles AE, Ellison WT, Finneran JJ, Gentry RL, Greene Jr CR, Kastak D, Ketten DR, Miller JH, Nachtigall PE. 2007. Marine mammal noise exposure criteria: initial scientific recommendations. Aquatic Mammals 33(4):411–509 DOI 10.1578/AM.33.4.2007.411. Southall BL, Scholik-Schlomer A. 2008. Potential application of vessel-quieting technology on large commercial vessels. In: Final report of the national oceanic and atmospheric administration (NOAA) international conference. Silver Spring: NOAA Fisheries. Stimpert AK. 2010. Non-song sound production and its behavioral context in humpback whales (Megaptera novaeangliae). PhD thesis, University of Hawaii. Szymanski MD, Bain DE, Kiehl K, Pennington S, Wong S, Henry KR. 1999. Killer whale (Orcinusorca) hearing: auditory brainstem response and behavioral audiograms. Journal of the Acoustical Society of America 106(2):1134–1141 DOI 10.1121/1.427121. Tasker ML, Amundin M, Andre M, Hawkins A, Lang W, Merck T, Scholik-Schlomer A, Teilmann J, Thomsen F, Werner S, Zakharia M. 2010. Marine strategy framework directive (task group 11 report) underwater noise and other forms of energy. Report

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 34/35 050 No. EUR, 24341. Available at http://ec.europa.eu/environment/marine/pdf/10-Task- Group-11.pdf . TC43 Acoustics. 2012. Acoustics. Quantities and procedures for description and measurement of underwater sound from ships. Part 1: general requirements for measurements in deep water. In: Publicly Available Specification 17208-1. Geneva: International Organization for Standardization. Thomsen F, Franck D, Ford JKB. 2000. Characteristics of whistles from the acoustic repertoire of resident killer whales (Orcinus orca) off Vancouver Island, British Columbia. Journal of the Acoustical Society of America 109(3):1240–1246. Tougaard J, Wright AJ, Madsen PT. 2014. Cetacean noise criteria revisited in the light of proposed exposure limits for harbour porpoises. Marine Pollution Bulletin 90(1–2) 196–208 DOI 10.1016/j.marpolbul.2014. Tyack PL. 2008. Implications for marine mammals of large-scale changes in the marine acoustic environment. Journal of Mammalogy 89(3):549–558 DOI 10.1644/07-MAMM-S-307R.1. UNCTD. 2013. Review of maritime transport. New York: United Nations Conference of Trade and Development. Urick RJ. 1983. Principles of underwater sound. Vol. 3. New York: McGraw-Hill. Veirs V, Veirs S. 2006. Average levels and power spectra of ambient sound in the habitat of Southern Resident orcas. Washington, D.C.: NOAA/NMFS/NWFSC. Wagstaff RA. 1973. RANDI: research ambient noise directionality model. Technical Report, DTIC Document. Available at http://www.dtic.mil/dtic/tr/fulltext/u2/ 760692.pdf . Wales SC, Heitmeyer RM. 2002. An ensemble source spectra model for merchant ship- radiated noise. The Journal of the Acoustical Society of America 111(3):1211–1231 DOI 10.1121/1.1427355. Wenz GM. 1962. Acoustic ambient noise in the ocean: spectra and sources. Journal of the Acoustical Society of America 34(12):1936–1956 DOI 10.1121/1.1909155. Williams R, Erbe C, Ashe E, Beerman A, Smith J. 2014. Severity of killer whale behavioral responses to ship noise: a dose–response study. Marine Pollution Bulletin 79(1– 2):254–260 DOI 10.1016/j.marpolbul.2013.12.004.

Veirs et al. (2016), PeerJ, DOI 10.7717/peerj.1657 35/35 051 Annex 6.D.41

Fate of Photodegraded Diluted Bitumen in Seawater.

052 Chemical Fate of Photodegraded Diluted Bitumen in Seawater

Zeyu Yang, Bruce P. Hollebone, Gong Zhang, Carl E. Brown, Chun Yang, Patrick Lambert,

Zhendi Wang, Mike Landriault, and Keval Shah

Emergencies Science and Technology Section (ESTS)

Science and Technology Branch, Environment and Climate Change Canada

Ottawa, ON, Canada

E-mail address: zeyu.yang@canada; [email protected]; [email protected];

[email protected]; [email protected]; [email protected];

[email protected]; [email protected]; [email protected]

ABSTRACT2017-336:

Diluted bitumen (dilbit), an oil sands product, may present new response challenges differing from conventional crude oil in terms of its potential environmental impacts. Simple naphthenic acids (NAs), a complex group of monocarboxylic acids, with a general formula

CnH2n+zO2, may be present in the source bitumen or may be created by photolytic weathering.

Knowing the composition and concentrations of NAs created during the photo-degradation process of dilbit will help understand the fate, behavior and toxicity of dilbit.

In the present study, two diluted bitumen products, Cold Lake Blend (CLB) and Access

Western Blend (AWB), were mixed with saltwater and irradiated with natural solar light

(Ottawa, Canada, 45.4°N) over winter and summer seasons, to assess the impact of sunlight on the chemical fate of the dilbit. For comparison, a light, sweet crude oil was exposed under similar conditions. The samples were analyzed by high performance liquid chromatography-high resolution mass spectrometry to examine the molecular transformation of diluted bitumen by

053 solar irradiation. The abundances of NAs in all three test oils increased significantly after 90 days of solar irradiation, strongly suggesting that polar NAs were formed by photolysis. Further, greater increases in NAs in the light crude were found than in the two dilbits. Similarly, the lighter oil had higher photolytic removal rates of petroleum hydrocarbons than the two dilbits.

The concentrations of NAs in oils exposed during the summer were generally higher than those exposed in winter. During summer exposure, the abundance of total NAs increased up to the 30-day’s solar exposure, then fell again, indicating the transient nature of these compounds.

However, net increases in polar NA compounds were observed for all the winter exposed samples. Greater increases were observed in the smaller NA compounds (average C-number decreased), also accompanied by an increase in saturation (average z-number decreased).

These chemical changes strongly indicate the effect of sunlight on the potential behaviour, fate and effects of spilled oil, with creation of new resin group compounds and reduction of aromatics and saturates. These changes may affect the viscosity of the oil and its ability to uptake water. These chemical compositions also imply significant changes to the ecological effects of the oil following a spill when aged in sunlight.

INTRODUCTION:

Diluted bitumen (dilbit), is a mixture of 20−30% of a diluent (e.g. condensates) and bitumen (Crosby et al., 2013). Similar to conventional crude oil, dilbit is immediately subject to a variety of abiotic and biotic processes including evaporation, dispersion, photo-oxidation, and microbial degradation once spilled into the marine environment. Some recent studies found that the fate of spilled dilbit depends strongly on the nature of the spill. Their behavior is reported to be different from some conventional petroleum products (i.e. faster evaporation, more rapid

054 mixing with sediments, and sinking of oil residues than conventional oil) (Government of

Canada, 2013).

Among these processes, photochemical degradation mediated by sunlight is an important pathway that influences their transformation and fate in the environment. Recent GC-based studies in our laboratory (Yang et al., 2016) have demonstrated that the chemical structure of petroleum hydrocarbons controls the photolysis of different groups in two representative dilbits

(Accessed Western Blend (AWB) and Cold Lake Blend (CLB)) and a conventional light crude oil (Alberta Sweet Mixed Blend, ESTS pour #5, ASMB#5). Briefly, aromatic hydrocarbons were photo-oxidized most rapidly, followed by n-alkanes, then the biomarker steranes and terpanes.

However, oil properties, temperature and solar intensity are other factors affecting the photolysis rates of the petroleum hydrocarbons in different oils.

Like conventional crude oil, dilbit is a complex mixture comprised of saturates, aromatics, resins and asphaltenes. Classic naphthenic acids (NAs, or O2-NAs) represent complex mixtures of alkyl-substituted aliphatic and cyclic monocarboxylic acids, described by the general formula

CnH2n+zO2, where n is the number of carbon atoms in the molecule and z is a negative, even integer that specifies hydrogen deficiency due to the presence of ring structures (Clemente and

Fedorak, 2005; Brown and Ulrich, 2015). It is noted that only O2-NAs will be investigated in this study. If no specific definition is used, NAs in this study will represent O2-NAs only. Among polar components of petroleum containing heteroatoms, NAs with monocarboxylic function, suspected to be the primary contributors to total acid number (TAN) in crude oil, are the most abundant oxygen-containing components in crude oil and diluted bitumen (Colati et al., 2013), and major components correlated to toxicity (Yue et al., 2016).

055 Abiotic and biotic weathering processes of oil create oxygenated transformation products

(Aeppli et al., 2012; Islam et al., 2013; Lemkau et al., 2014; Ray et al., 2014; Folwell et al.,

2016). NAs, are one of the main classes of by-products of incomplete biodegradation, photodegradation or thermal degradation of petroleum products (Ray et al., 2014; Folwell et al., 2016).

Furthermore, the photo-oxidation of the non-polar petroleum hydrocarbons in dilbit may simultaneously result in the formation and the photo-oxidation of NAs. Beyond studying the alteration of petroleum hydrocarbons, the study of these polar oxygenated components subjected to photo-oxidation is another approach to understand the photo-oxidation mechanism of dilbit. It will help us fully understand the fate and behavior of spilled oil in the environment, especially those resulting from long-term weathering processes in the environment, containing abundant polar components compared to conventional crude oil and their refined products.

The present study is an extension of our earlier work (Yang et al., 2016). The variation of petroleum hydrocarbons, e.g., polycyclic aromatic hydrocarbons (PAHs) and their alkylated congeners (APAHs), total petroleum hydrocarbons (TPH) in diluted bitumen by solar irradiation has been investigated. This study will provide a molecular snapshot of the NA compositional changes at different solar exposure time points and seasons in dilbit. This information will help us to understand the fate of NAs present in dilbit or formed through photo-oxidation of dilbit in the environment. As a caution, it is important to note that as no certified reference materials for

NA compounds yet exist, all values reported in this study must be regarded as semi-quantitate only.

METHODS:

Chemical and reagents

056 LC/MS purity acetic acid, formic acid and ammonium acetate, solvents with HPLC grade

2 (methanol, isopropanol and acetonitrile), deuterated fatty acids ([ H15] caprylic acid (C8:0-d15),

2 2 [ H2] palmitic acid- (C16:0-d2), [ H27] myristic acid (C14:0-d27), and 13 straight chain saturated fatty acids (SSFA) from C6 to C30 (even number only), were supplied by Sigma-Aldrich (St.

Louis, MO, USA). All other solvents used were the highest purity (Caledon, Canada). Ultra pure water was prepared from a Milli-Q water purification system (Millipore, Billerica, MA, USA).

Silica gel (100−200 mesh) was supplied by Spectrum Chemicals (Gardena, CA, USA). Purified

Merichem NA mixture was provided by the Pacific and Yukon Laboratory for Environmental

Testing (PYLET), Environment and Climate Change Canada.

Exposure experiments

Dilbits including CLB and AWB, and one representative conventional light crude oil

ASMB#5) were diluted by hexane with a small amount of dichloromethane (DCM) to 150 mg/mL. The detailed physicochemical information for the oils used in the present study was presented in the reference (Yang et al., 2016). Briefly, the two dilbits, especially AWB, have higher abundances of resins and asphaltenes than ASMB#5,. ASMB#5 contains a higher content of resolved peaks (35% of total petroleum hydrocarbons), as well as less unresolved complex mixture (UCM) than the two dilbits. The characteristic of the chemical pattern suggests that the two dilbits contain more of the heavier recalcitrant compounds than the ASMB#5 crude oil.

The solar exposure process for the oil-water samples was described in the reference

(Yang et al., 2016). In brief, 300 mg of oil was spiked onto the surface of artificial seawater. The samples were covered with a polyethylene membrane/aluminium foil after the solvent had evaporated to provide the protection from the rain and other materials, and minimize evaporation loss during solar exposure. Then the samples were placed on the roof top of a 2-story building in

057 Ottawa for winter exposure from December 18, 2014 to March 19, 2015, and summer exposure from May 13, 2015 to September 11, 2015.

For every test, three samples were prepared. Two samples were covered with polyethylene membrane to minimize evaporation loss while exposed to solar light; the third sample was covered with aluminum foil and kept in the same environment as a control sample to eliminate the uncertainties caused by microbial degradation and temperature differences during sunlight exposure. All control and test samples were taken for further analysis after 5, 15, 30, 60, and 90 days of solar exposure, respectively,.

Analytical procedures

All solar exposed and control water samples were acidified with hydrochloric acid to a pH value of ≤2 and spiked with the appropriate surrogates. Then they were extracted with DCM three times. The combined liquid-liquid extraction (LLE) extracts were concentrated to 10 mL for the following analytical procedures. Appropriate extracts were loaded into a column (5 mm i.d.) with 1 g silica gel and topped with 1 cm of anhydrous Na2SO4. Ten mL of DCM was used to remove the non-targeted petroleum hydrocarbons to minimize instrumental contamination.

Another 7 mL of methanol containing 0.1% formic acid was used to elute the polar targets. The elutes were then dried with a gentle stream of nitrogen gas, and reconstituted in 1 mL of HPLC grade isopropanol with 1.0 µg/mL of deuterated palmatic acid (d2-C16:0) as an internal standard.

All samples were centrifuged at 3000 rpm for 10 min prior to HPLC-HRMS analysis.

HPLC-HRMS analysis of acidic extracts

Sample analysis was performed on an Accela HPLC system paired with Exactive

Orbitrap Mass System (Thermo Fisher Scientific, San Jose, CA, USA) with electrospray

058 ionization (ESI) source in negative ion mode. A 20 min LC gradient method using a Poroshell

120 EC-C8 column (100 × 2.1 mm i.d., 2.7 µm, Agilent) at a flow rate of 0.25 mL/min was used for separation. The gradient method was programmed among water with 0.0025% of formic acid

(A); and 95%/5% acetonitrile/water with 3 mmol/L ammonium acetate (B) , 85%/10%/5% isopropanol/toluene/water with 3 mmol/L ammonium acetate (C). In detail, 95% mobile phase A and 5% mobile phase C held isocratically for 2 min, followed by the linear gradients from 5% to

100% mobile phase B in 8 min. Then mobile phase B decreased to 10% and C increased to 90% linearly in the following 5 min, and held for 10 min. After that, mobile phase b increased to

100% in 0.5 min, held for 1.5 min, and then returned to 5% B and 95% A in 0.5 min, and held for 5 min prior to the next injection. The Orbitrap mass spectrometer was operated in ESI negative mode (ESI-) and data was acquired in full scan from the mass to charge (m/z) ratio of

80 to 1600. The negative ion of dimmer of acetic acid (2M-1, m/z=119.03498) was used as a mass-lock for scan-to-scan calibration correction to get less than 2 ppm mass accuracy. The mass parameters were set as follows: spray voltage, 4.0 kv; capillary temperature, 320oC; sheath gas,

45; auxiliary gas, 15; and tube lens, -90 v.

Heteroatom containing species including the formula of CnH2n+zO2 were detected as

- deprotonated molecules by the Orbitrap MS ESI-scan, that are [M-H] ions, while only the O2-

NAs were identified based on the molecular weights, and quantified by the following model standards in the present study. The MS data were processed, and the elemental compositions of the compounds were determined by measuring the accurate m/z values. Mass accuracy was set at

±5 ppm. Concentrations of NAs spanning from C6 to C60 with z=0 to -16 were calibrated based on the average response factor (RRF) of 1 mg/L of SSFA standards. Mass spectra were acquired and processed using the Xcalibur software package (Thermo Fisher Scientific, San Jose, CA,

059 USA). To help visualize and interpret the MS data, the typical 2-dimensional and 3- dimensional plots were constructed, such as a carbon number or z number versus intensity, or carbon number versus z number, where z number specifies hydrogen deficiency.

RESULTS/DISCUSSION:

NAs in the control oils

Plots of the z number versus the carbon numbers have proven to be useful tools for the differentiation of complex organic mixtures based on chemical composition (Yue et al., 2016).

These plots allow the presentation of all signals from a specific class in a simple and feasible way. To quantitatively compare the NAs in the three test oils, carbon versus z number plots were generated for the three control oils (Figure 1). The two dilbits (CLB and AWB) have higher intensity and wider distribution of NAs than ASMB#5. The intensity of NAs with different z number generally decreases with the increase of the z numbers for the three oils, except for the differences noted below. Specifically, NAs with z=0, -2, and -4, typical NAs of 0−2 naphthenic rings, are the most abundant groups in CLB, followed by z=-6, then the others with higher number of naphthenic rings. In AWB, NAs with z=-2, -4 and -6, typical NAs with 1−3 naphthenic rings, are the most abundant groups, followed by z=0, -8, and -10, and then the others with higher number of naphthenic rings. In ASMB#5, acyclic NAs (z = 0) are the most abundant group, followed by z= -2, and -4 (NAs of 1−2 naphthenic rings), and then the others. The analysis of the distribution profiles of NAs by carbon number demonstrates that a typical bell shape was observed from the carbon number of 6 to 60. The intensity of light or heavy NAs is relatively low compared to the medium ones. Specifically, C16−C30 NAs are the most abundant components in CLB and AWB, while some light NAs are also abundant in ASMB#5. The most

060 abundant NA components in ASMB#5 range from C15 to C23; the distribution of NAs in

ASMB#5 in the C6−C60 range is much narrower than for the two dilbits. ASMB#5 is a representative of a light crude oil, while CLB and AWB represent heavy diluted bitumen products in the present study. ASMB#5 has less polar components (resins and asphaltenes) than the two dilbits (Yang et al., 2016) because a high level of polar oxygenated intermediates have been formed through the long term natural degradation of bitumen in the dilbits.

Variation of total NAs through photo-oxidation

The variation of total quantified NAs with solar exposure time in the two seasons is depicted in Figure 2. In summer, all the three test oils show a rapid increasing trend from the control samples to 15 days solar exposure; then a slow increasing trend that lasts until 30 days of exposure; then a decreasing trend can be observed from 60 to 90 days exposure. This trend is similar to the photolysis of APAHs or total petroleum hydrocarbons (Yang et al., 2016). NAs were formed rapidly at the beginning of solar exposure for all the three test oils due to the photo- oxidation of some non-polar petroleum hydrocarbons, especially for aromatic hydrocarbons

(Yang et al., 2016). The decreasing trend after 30 days exposure suggests that NAs present in the exposed system have been photo-oxidized simultaneously due to the presence of photo-sensitive components. The photo-sensitive non-polar petroleum hydrocarbons were essentially depleted after 30 days summer exposure. In this specific scenario, the formation rates of NAs were less than their photolytic rates because the newly formed NAs could not replace the photo-oxidized congeners, which are expressed in the decreased total NAs.

In winter exposure, an increasing trend can be observed for both CLB and ASMB#5 during the 90 days exposure. It is noted that AWB was not tested in winter season. As noted in reference (Yang et al., 2016), the photo-oxidation of petroleum hydrocarbons relies on the solar

061 intensity, as well as the environmental temperature. It is reasonable because higher solar intensity and temperature has resulted in petroleum hydrocarbons being photo-oxidized faster in summer than in winter. In winter exposure, the replacing of NAs are continuous because the maximal final photolytic rate of total aromatic hydrocarbons in ASMB#5 is 92%, while it is 72% in winter exposed ASMB#5 (Yang et al., 2016). This further confirms that solar intensity and temperature are essential for the photo-oxidation of oil components, including the previously reported petroleum hydrocarbons (Yang et al., 2016) and polar oxygenated components.

The comparison of the two dilbits and ASMB#5 indicates that NAs were formed faster in

ASMB#5 than the two dilbits, although the NAs are very low in the ASMB#5 control. It is clear that the fast photo-oxidation rates of ASMB#5 have resulted in more NAs being formed than in the two dilbits. This conclusion is supported by the our previous results (Yang et al., 2016), where we concluded that the initial oil chemical composition is another factor contributing to the photo-oxidation rates of petroleum hydrocarbons.

Variation of NAs with different carbon range

The concentrations for all NAs with same carbon number but different z numbers were added together to evaluate the variation of NAs with carbon range (C6−C60) through the photolytic process. Figure 3 displays the concentration of NAs versus different carbon number for the control and solar exposed test samples. Abundant NAs with different carbon numbers

(from C6 to C60), especially in the carbon range of C6−C16 (5−10 times that of the control oils), were formed through photo-oxidation in the two seasons. Similar to the total NAs, the highest abundance of the sub-summed NAs in the three oils can be observed after 30 days exposure in summer, while the NAs with different carbon numbers after 60 and 90 days summer exposure are generally lower than after 15 days of exposure. Therefore, some newly formed or pre-

062 existing NAs were photo-oxidized simultaneously after 30-day summer exposure. An increasing tendency can be found during the 90 days winter solar exposure. The most abundant NAs in

CLB and AWB located at C21 in the control samples; it shifted to C16 after solar irradiation in the two seasons. Similarly, the most abundant NAs located at C18 in ASMB#5 control samples, while it shifted to C16 after solar exposure. Except the maximal peak shifted to the smaller carbon number, abundant NAs in the range of C7 to C16 were formed in the three test oils after solar exposure. The abundances of these NAs are usually higher than those with carbon number greater than 45. Accordingly, abundant light molecular NAs were formed during photo-oxidation.

This finding is similar to earlier studies, where a shift towards lower carbon numbers after the photo-catalytic treatment of acidic extracted organics in OSPW was observed for both the O2 and

O4 NA class. It was also reported that the toxicity of the water phase was enhanced significantly after photo-oxidation (Maki et al., 2001), because more water soluble polar oxygenated intermediates were dissolved in water phase after photo-oxidation (Ray et al., 2014). The left shift of these O2 NA class in the present study can be ascribed to the preferential photo-oxidation of higher molecular weight acidic extracted organics (Leshuk et al., 2016). The finding for the preferential photo-oxidation of higher molecular weight acidic extracted organics may represent another oil spill treatment option, because NAs with higher carbon numbers were reported to be the most environmentally persistent and resistant to biodegradation (Misiti et al., 2014).

Furthermore, this similarity suggests that the photo-oxidation of NAs is independent on the sample matrix, either in OSPW or in water contaminated with oil. However, the photo-oxidation of petroleum hydrocarbons (e.g., aromatic hydrocarbons, or some of the saturated compounds) is preferential in the water matrix with abundant oil as observed with the increasing trend of NAs in

Figures 2 and 3.

063 For the different oils studied, (except for the summer exposed AWB), the concentrations of sub-summed NAs in all the exposed samples are higher than their respective control samples in both seasons, a decreasing trend was observed after 30 days in the summer exposed CLB and

ASMB#5. This phenomenon suggests the photo-oxidation of non-polar petroleum hydrocarbons has formed abundant NAs as intermediates. The increased NAs also can be supported by the increased acidity for the water phase, where saltwater was spiked with dilbit on the surface and then exposed to the natural sunlight (Government of Canada, 2013). The abundances of NAs with carbon number > 20 in AWB are less than the corresponding control samples after 30−90- day summer exposure. In this scenario, the photolysis rates of these NAs are faster than their formation rates. Additionally, the relatively increased level of the newly formed NAs varies with oil type and carbon chain length. Specifically, it is limited in AWB compared to CLB and

ASMB#5, consistent with the above observations. Therefore, the variation of NAs for a given oil is specific during the photolytic process, depending on individual oil property.

Relative production of NAs with as a function of aromaticity (z number)

Similarly, the concentrations for all NAs with the same z number but different carbon number were added together to evaluate the variation of NAs with z numbers (0−-16, corresponding double bond equivalence (DBE) from 0 to 8, and naphthenic rings from 0−8) through the whole photolytic process (Figure 4). In the photo-oxidized samples, the distribution profiles of the sub-summed NAs are almost the same for all the samples in the z number range of

0 to -16. For example, they generally decrease with increasing of z number, except that some of the z=-8 NAs are higher than the z=-6 congeners in ASMB#5 in the both seasons. That is to say, the most abundant NAs with z=0 were formed through the photolysis of dilbits and ASMB#5, followed by a generally decreasing trend with the increase of the negative z numbers. The

064 formation of NAs with different carbon and z number can be theorized to be by preferential photo-oxidation of aliphatic, cyclic or aromatic hydrocarbons to aldehydes, or ketones (O1), further hydroxylated carboxylic acid (O2) intermediates, and even ring opening for aromatic intermediates. Some evidence for these compounds has been provided by negative-ion ESI (Yue et al., 2016). However, the decreased concentrations of NAs with the increased naphthenic rings may be correlated to the photo-sensitivity of the NAs themselves. We must assume that the formation and photolysis of NAs occurred simultaneously during the solar exposure process.

NAs with more naphthenic rings (double bonds) may be more photo-sensitive, which could result in their faster photo-degradation than those with less double bonds. The exception of more

NAs with z=-8 formed in ASMB#5 than CLB and AWB, may suggest more NAs with 4- naphthenic ring were formed in ASMB#5 than 3-ring congeners. This may indicate that they are sourced from the photo-oxidation of aromatic hydrocarbons, because they have the highest photolytic rates in both seasons for the ASMB#5 crude oil.

The variation of the sub-summed NAs with the exposure time also depends on exposure season too. All of the sub-summed NAs increase significantly until 15−30 days of summer solar exposure; a decreasing tendency can be observed with the continued exposure for all the three test oils, especially for ASMB#5. It is noted that NAs with z number =-14 and -16 in ASMB#5 begin to decrease after 15 days solar exposure, indicating NAs with more cyclic or unsaturated compounds have higher photo-reactivity (Leshuk et al., 2016). It is possible that these O2 NAs are further oxidized to O3−O8 species during exposure (Ray et al., 2014; Vaughan et al., 2016). It is a promising finding because acidic extracted organic species with more cyclic rings are the most environmental persistent and resistant to biodegradation (Misiti et al., 2014), which

065 suggests photo-oxidation of these higher priority targets may be a potential alternative to oil spill treatment.

In winter exposed samples, a general increasing tendency for all of the sub-summed NAs was observed until 90 days for ASMB#5. In CLB, a significant increase for NAs with z= 0 was observed for the first 60 days, while all the other z numbers were essentially unaltered; an abrupt increase was found for those with z numbers of -2, -4, -6, -8, and -10 after 90 days of exposure.

This is reasonable because the formation of NAs is faster than their loss by photolysis at the beginning of summer or the whole winter exposure, which expresses as an increased level of

NAs. However, the continual summer exposure could not replace the loss of NAs, because of the depletive exhaustion of petroleum hydrocarbons, especially in ASMB#5, which has the highest photolytic rates among the three oils.

CONCLUSIONS:

The O2 NAs in oils exposed to natural sunlight during the summer were generally more abundant than those exposed in winter. During summer exposure, the total NAs abundances rose until the 30-day mark, then dropped again, indicating the transient nature of these compounds.

However, net increases in polar NA compounds were observed for all the winter exposed samples. Greater increases were observed in the smaller NA compounds (average C-number decreased), accompanied by an increase in saturation (average z-number decreased).

ACKNOWLEDGEMENTS:

This work was funded and supported by the Government of Canada’s World Class Tanker Safety

System (WCTSS) program.

066

CLB control

200

)

( 150

a 100

t n

e 60

c 50

n 50 er

o 40 b C m 0 30 nu 0 20 n -2 -4 o -6 -8 10 rb -10 -12 a -14 -16 0 C z -18 number

AWB control

500

)

g /

g 400

(

o 300 i

a r

t 200 n

e 60 c

n 100 50 er

o 40 b C m 0 30 nu 0 20 n -2 -4 o -6 -8 10 rb -10 -12 a -14 -16 0 C z -18 number ASMB#5 control

)

g /

g 50

(

n 40

a 30

r t

n 20

e 60 c

n 10 50 er

o 40 b C m 0 30 nu 0 20 n -2 -4 o -6 -8 10 rb -10 -12 a -14 -16 0 C z -18 number

Figure 1 Comparison of the distribution profiles of naphthenic acids in the three test oils as

controls

067 60000 ASMB#5 CLB Summer 50000 AWB

40000

30000

20000

10000 g/g) µ 0

Control 5 15 30 60 90 50000 ASMB#5 CLB Winter Concentration ( 40000

30000

20000

10000

Control 5 15 30 60 90

Solar exposure time (days)

Figure 2 Variation of total naphthenic acids in different seasons

068 (a) 6000 ASMB#5 5000 Control 5 d 4000 15 d 3000 30 d 60 d 2000 90 d 1000 0

4000 CLB 3000 g/g) µ 2000

1000

4000 AWB Concentration ( 3000

2000

1000

(b) 4000 ASMB#5 Control 3000 5 d 15 d 2000 30 d 60 d 1000 90 d g/g) µ 0

4000 CLB 3000

2000

Concentration ( 1000

10 15 20 25 30 35 40 45 50 55 60

NAs with different carbon number Figure 3 The variation of NAs with different carbon chain length in different exposure

season. (a), summer; (b), winter.

069 (a)

25000 ASMB#5 Control 20000 5 d 15 d 15000 30 d 60 d 10000 90 d

5000

0 16000 CLB 12000

g/g) 8000 µ

4000

0 16000 AWB 12000 Concentration ( 8000

4000

0 (b) 12000 ASMB#5 Control 5 d 8000 15 d 30 d 60 d 90 d

g/g) 4000 µ

0 16000 CLB 12000

8000 Concentration (

4000

0 Z=0 Z=-2 Z=-4 Z=-6 Z=-8 Z=-10 Z=-12 Z=-14 Z=-16

NAs with diffferent z numbers Figure 4 The variation of the concentration of NAs with different z number in different exposure seasons, (a), summer; (b), winter.

REFERENCES:

070 Aeppli, C., Carmichael, C.A., Nelson, R.K., Lemkau, K.L., Graham, W.M., Redmond, M.C., Valentine, D.L., Reddy, C.M. 2012. Oil weathering after the Deepwater Horizon disaster led to the formation of oxygenated residues. . Environmental Science & Technology. 46: 8799-8807.

Brown, L.D., Ulrich, A.C. 2015. Oil sands naphthenic acids: A review of properties, measurement, and treatment. Chemosphere. 127: 276-290.

Clemente, J.S., Fedorak, P.M. 2005. A review of the occurrence, analyses, toxicity and biodegradation of naphthenic acids. Chemosphere. 60: 585-600.

Colati, K.A.P., Dalmaschio, G.P., de Castro, E.V.R., Gomes, A.O., Vaz, B.G., RomÃ£o, W. 2013. Monitoring the liquid/liquid extraction of naphthenic acids in brazilian crude oil using electrospray ionization FT-ICR mass spectrometry (ESI FT-ICR MS). Fuel. 108: 647-655.

Crosby, S., Fay, R., Groark, C., Kani, A., Smith, J.R., Sulivan, T., Pavia, R. 2013. Transporting Alberta oil sands products: defining the issues and assessing the risks. Seattle, WA.

Folwell, B.D., McGenity, T.J., Price, A., Johnson, R.J., Whitby, C. 2016. Exploring the capacity for anaerobic biodegradation of polycyclic aromatic hydrocarbons and naphthenic acids by microbes from oil-sands-process-affected waters. International Biodeterioration & Biodegradation. 108: 214-221.

Government of Canada. 2013. Properties, composition, and marine spill behaviour, fate and transport of two diluted bitumen products form the Canadian oil sands. , accessed at November 20, 2016.

Islam, A., Cho, Y., Yim, U.H., Shim, W.J., Kim, Y.H., Kim, S. 2013. The comparison of naturally weathered oil and artificially photo-degraded oil at the molecular level by a combination of SARA fractionation and FT-ICR MS. Journal of Hazardous Materials. 263, Part 2: 404-411.

Lemkau, K.L., McKenna, A., Podgorski, D.C., Rodgers, R.P., Reddy, C.M. 2014. Molecular evidence of heavy-oil weathering following the M/V Cosco Busan spill: Insights from Fourier

071 Transform Ion Cyclotron Resonance Mass Spectrometry. Environmental Science & Technology. 48: 3760-3767.

Leshuk, T., Wong, T., Linley, S., Peru, K.M., Headley, J.V., Gu, F. 2016. Solar photocatalytic degradation of naphthenic acids in oil sands process-affected water. Chemosphere. 144: 1854- 1861.

Maki, H., Sasaki, T., Harayama, S. 2001. Photo-oxidation of biodegraded crude oil and toxicity of the photo-oxidized products. Chemosphere. 44: 1145-1151.

Misiti, T.M., Tezel, U., Pavlostathis, S.G. 2014. Effect of alkyl side chain location and cyclicity on the aerobic biotransformation of naphthenic aicds. Environmental Science & Technology. 45: 7431-7437.

Ray, P.Z., Chen, H., Podgorski, D.C., McKenna, A.M., Tarr, M.A. 2014. Sunlight creates oxygenated species in water-soluble fractions of Deepwater horizon oil. Journal of Hazardous Materials. 280: 636-643.

Vaughan, P.P., Wilson, T., Kamerman, R., Hagy, M.E., McKenna, A., Chen, H., Jeffrey, W.H. 2016. Photochemical changes in water accommodated fractions of MC252 and surrogate oil created during solar exposure as determined by FT-ICR MS. Marine Pollution Bulletin. 104: 262-268.

Yang, Z., Hollebone, B.P., Brown, C.E., Yang, C., Wang, Z., Zhang, G., Lambert, P., Landriault, M., Shah, K. 2016. The photolytic behavior of diluted bitumen in simulated seawater by exposed to the natural sunlight. Fuel. 186: 128-139.

Yue, S., Ramsay, B.A., Wang, J., Ramsay, J.A. 2016. Biodegradation and detoxification of naphthenic acids in oil sands process affected waters. Science of the Total Environment. 572: 273-279.

072 415

Fate of Oxygenated Components for Solar lrradiated Diluted Bitumen in Saltwater

ZeyuYang, Gong Zhang, Bruce P. Hollebone, Carl E. Brown, Chun Y*g, Patrick Lambert, Mike Landriault, and Keval Shah Emergencies Science and Technology Section Science and Technology Branch, Environment and Climate Change Canada Ottawa, ON, Canada zevu.yang(ùcanada.ca

Abstract Diluted bitumen was mixed with saltwater and irradiated with natural solar light in Ottawa, to assess the impact of sunlight on its fate. For comparison, a crude oil (ASMB#5) was exposed in sunlight under same situation to evaluate the similarities and differences between non- conventional and conventional crude oils. The oxygenated components, including carbonyl polycyclic aromatic hydrocarbons (PAHs), and acidic polar fractions (the naphthenic acid fraction compounds, NAFCs) with the elemental formulae of C.H¡N'O.S, (where c, h, n, o, and s are the number of carbon, hydrogen, nitrogen, oxygen and sulfur in the following range, 6 < c < 100, 6

Yang,Z., G. Zhang, B.P. Hollebone, C.E. Brown, C. Yang, P.G. Lambert, M. Landriault, and K. Shah, Fate of Oxygenated Components for Solar lrradiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar, Environment and Climate Chanoe Canada. Ottawa. ON. oo. 4'15-440.2017.

073 416

controlled the photolysis of different groups in dilbits and a conventional light crude oil (Alberta Sweet Mixed Blend #5, ASMB#5). Specifically, aromatic hydrocarbons were photo-oxidized fastest, followed by n-alkarres, then the weathering-resistant biomarker steranes and terpanes. Simultaneously, oil properties, temperature and solar intensity are the other factors affecting the photolysis rates of the petroleum hydrocarbons. It has been reported that abiotic and biotic weathering processes create oxygenated transformation products accounting for 60-80% of extractable material (Aeppli et aI.,2012). Similarly, it has been reported that photo-oxidation may most likely enhance the dispersion of the oil spills by transforming the oil components into more water-soluble oxygenated ones (Ray et a1.,2014; Wang el al,2016). Studies have concluded that photo-oxidation products exert more adverse effects on algae, bacteria, and marine invertebrates and fish (Maki et a1.,2001; Halton et al., 2010). Some researchers also reported that huge discrepancy existed between the abundance of the conventional GC detected petroleum hydrocarbons (e.g., PAHs) and biological responses, which suggested that those unidentif,red remains were responsible for the adverse biological responses, not PAHs only (Halton et al., 2010; Incardona et al., 2013). Accordingly, these newly formed oxygenated photoproducts have greater toxic effects on marine life and environment than the petroleum hydrocarbons present in the original oil. It is well known that GC-based techniques are limited to compounds with volatility oC. below around 400 Therefore, some petroleum hydrocarbons (e.g., the GC-detectable petroleum hydrocarbons, n-alkanes, polycyclic aromatic hydrocarbons (PAHs) and their alkylated congeners (APAHs), and some petroleum biomarkers) can be analyzedby them (D'Auria et al., 2008 and 2009; John et a1.,2016; Yang et a1.,2016). Except for the GC-detected targets, most of the newly formed oxygenated intermediates through biodegradation/photo- oxidation generally have low volatility and high polarity due to the addition of oxygen. Therefore, the identification of the components with higher polarity, water solubility and non- volatility will help us fully understand the fate and behavior of spilled oil in the environment. It is crucial to identify these transformed molecules occurring to petroleum hydrocarbons after environmental spill, especially for those suffered from long-term weathering processes in the environment. One class of the photoproducts is the oxygenated PAHs, especially for PAH ketones and PAH quinones, which are formed through the photo transformation of PAHs as the intermediates for further minerclization or degradation. They are even more toxic than the parent compounds (Mallakin et a1.,1999). The other major polar photoproducts are the naphthenic acid fraction compounds (NAFCs) containing diverse family of carboxylic acids and other acid-extractable organic compounds with varied structures (e.g., aromatic, adamantane or diamondoid structures) and different species (such as sulfur- and nitrogen-containing compounds and multiple oxygenated acids) (Rowland et a1.,201 1). NAFCs have been reported to be the principal toxicants in oil sands processed waters (OSPW) (Clemente and Fedorak,2005; Frank et a1.,2008; Frank et a1.,2009), while the conventional identification technology is limited in identifying and quantifying them due to their complex composition. Currently, some advanced analytical techniques, such as Fouier Transform Ion Cyclotron Resonance mass spectrometry (FT-ICR MS), have been used to characterize those polar and nonvolatile compounds through photochemical transformation (Headley eTal.,2011; Islam et a1., 2013; Ray et a1.,2014; Vaughan et al., 2016), biodegradation (Hughey et al., 2007), naturally weathering (Lemkau et al., 2014), or in some crude oils (Stanford et al., 2007). Ultrahigh resolution electrospray ionization in negative mode (ESI-) Orbitrap mass spectrometry @SI(-) Orbitrap MS) was shown to be well-

Yang,Z., G. Zhang, B.P. Hollebone, C.E. Brown, C. Yang, P.G. Lambert, M. Landriault, and K. Shah, Fate of Oxygenated Components fo¡ Solar lrradiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar, Environment and Climate Chanoe Canada. Ottawa. ON, oo. 415-440.2017.

074 417

suited for the characterization of naphthenic acids with monocarboxylic acid (Ì.{As) and NAFCs in Athabasca basin OSPW(Headley et a1.,2012). It can therefore satisfy this purpose to identify photochemical transformations occurred to petrogenic molecules at the molecular level. The present study is an extension of our earlier work (Yan g et a1,2016). Two diluted bitumen oils, Cold Lake Blend (CLB) and Accessed Western Blend (AWB), were mixed with saltwater and inadiated with natural solar light (Ottawa, Canada,45.4'N) over winter and summer seasons, to assess the impact of sunlight on the chemical fate of the dilbit. For comparison, a light, sweet crude oil was exposed under similar conditions. The variation of some GC-detectable petroleum hydrocarbons, e.g., n-alkanes, PAHs, APAHs, and total petroleum hydrocarbons (TPH), in diluted bitumen by solar irradiation has been investigated in the earlier report (Yang et a1.,2016).In the present study, the samples at pre-determined time points were collected and gone through appropriate pre-treatment procedures, analyzed by GC-MS for carbonyl PAHs, and high performance liquid chromatography-ultrahigh resolution Orbitrap mass spectrometry combined with electrospray ionization in negative mode (HPLC-HRMS) for NAFCs. The variation of the chemical composition of representative carbonyl PAHs and NAFCs in diluted bitumen by solar irradiation was investigated. This study provides a molecular snapshot at the compositional changes of the oxygenated components at different solar exposure time points and seasons in dilbits, help us understand the fate of the oxygenated intermediates to evaluate the potential toxicity changes through photo oxidation. It is noted that no certified reference materials for identiffing and quantifying NAFCs yet exist because their chemical composition are extremely complex. The average mass spectra were generated automatically to identifu NAFCs based on their elemental composition for each sample. Therefore, all values reported in this study must be regarded as semi-quantization only because the reported results are the relative abundance normalized to the response of internal standard in each injection.

2. Methods 2.1. Chemical and Reagents LCIMS purity of acetic acid, formic acid and ammonium acetate, solvents with HPLC grade (methanol, isopropanol and acetonitrile), deuterated carbonyl PAHs ç¡2n61t,+- naphthequinone (1,4-naphthoquinone-d6), and 12Hs] 9-fluorenone (9-fluorenone-ds)), deuterated fatty acids (l'Hrtlcaprylic acid (C8:0-üs),12Hzlpalmitic acid- (C16:0-d2), ['HrrTmyristic acid (Cl4:}-du), 13 carbonyl PAH standards including l-indanone,I,4-naphthoquinone, acenaphthenequinone, 9-fluorenone, peinaphthenone, anthraquinone, 9, 1O-phenanthrenequinone, 2-ethyl anthraquinone, benzo[a]fluorenone, and 5,l2-nanphthacenequinone, and 13 straight chain saturated fatty acids (SSFA) from C6 to Go (even number only), were supplied by Sigma- Aldrich (Bellefonte, PA, USA).All other solvents used were the highest purity (Caledon, Canada). Ultra-pure water was prepared from a Milli-Q water purification system (Millipore, Billerica, MA, USA). Silica gel (100-200 mesh) was supplied by Spectrum Chemicals (Gardena, CA, USA). Purified Merichem NA mixture was provided by the Pacific and Yukon Laboratory from Environmental Testing (PYLET), Environment and Climate Change Canada.

2.2. ExposureExperiments Dilbits including CLB and AV/B, and one representative crude oil (ASMB#5) were diluted by hexane with a small amount of dichloromethane (DCM) to 150 mg/ml. It is noted that the dilbit products used in the present study are relatively viscous. Only 300 mg of oil was used for photolysis study. The dilution is crucial for quantitatively spiking oil into water phase to further quantify their photo-oxidation rates. Hexane is the preferred solvent, as it is supposed to

Yang,2., G. Zhang, B.P. Hollebone, C.E. Brown, C. Yang, P.G. Lambert, M. Landriault, and K. Shah, Fate of Oxygenated Components for Solar lnadiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar, Environment and Climate Chanoe Canada. Ottawa. ON. oo. 415-440.2017.

075 418

float on the surface of water, which lets hexane easily being evaporated prior to solar irradiation. DCM was spiked for CLB and AWB to facilitate their dissolution and homogeneity in the final stock solution. The detailed physicochemical information for the used oils was presented in the reference (Yang et al., 2016). Briefly, the two dilbits have higher resins and asphaltenes than ASMB#5, especially for AWB. ASMB#5 contains a higher content of resolved peaks (35% of total petroleum hydrocarbons), but has less unresolved complex mixture (UCM) than the two diibits. The characteristic of chemical pattern suggests that the two dilbits contain more of the heavier recalcitrant compounds than the ASMB#5. The solar exposure process for the oil-water samples was described in the reference (Yang et al., 2016).In briet 300 mg of oil was spiked onto the surface of artificial seawater. The samples were wrapped with a polyethylene membrane/aluminium foil after the solvent had evaporated to provide the protection from the rain and other materials, and minimize evaporation loss during solar exposure. Then the samples were placed on the roof top of a 2-story building in Ottawa for winter exposure from December 18, 2014 fo March 19,2015, and summer exposure from May 13, 2015 to September 1 1 , 2015 . It is noted that snow was removed timely to expose the samples to daylight in the winter exposure. For every test, three samples were prepared. Two samples were covered with polyethylene membrane to minimize evaporation loss while exposed to solar light; the third sample was fully wrapped with aluminum foil and kept in the same environment as a control sample to eliminate the uncertainties caused by microbial degradation and temperature differences during sunlight exposure. All control and test samples were taken for fuither analysis after 5, 15, 30, 60, and 90 days of solar exposure, respectively. All the reported data are the average data from the replicate samples, where each sample was analyzed twice by GC-MS or HRLC-MS. The error bars reported in this study are the standard derivation of the reported averages.

2.3. AnalyticalProcedures All solar exposed and control water samples were acidified with hydrochloric acid to a pH value of <2 and spiked with the appropriate surrogates (Yang et a1.,2016). Then they were extracted with dichloromethane (DCM) three times. The combined liquid-liquid extraction (LLE) extracts were concentrated to 10 mL for the following analytical procedures. For carbonyl PAH analysis, appropriate extracts were loaded into a column (10 mm i.d.) with 3 g silica gel and top with I cm of anhydrous NazSO¿. Following the elution of aliphatics and aromatics with hexane and the mixture of hexane and DCM (1:1, v: v), 15 mL of the mixture of DCM and acetone (95:5 of v: v) was used to elute the carbonyl PAHs. The elutes were then dried by gentle N2, spiked with d1a-terphenyl as internal standard for carbonyl PAHs analysis. For the analysis of NAFCs, appropriate extracts were loaded into a column (5 mm i.d.) with I g silica gel and top with 1 cm of anhydrous NazSO¿. Ten mL of DCM was used to remove the interferences from the target components. Another 7 mL of methanol containing 0.1% formic acid was used to elute the polar targets out. The elutes were then dried by gentle N2, and reconstituted in I mL of HPLC grade isopropanol with 1.0 ¡rg/ml of deuterated palmatic acid (d2-CI6:0) as internal standard. All samples were centrifuged at 3000 rpm for 10 min prior to HPLC-HRMS analysis.

2.4. InstrumentalAnalysis 2.4.1. GC-MS Analysis of Carbonyl PAHs Characterizations of carbonyl PAHs were performed on an Agilent 6890 GC system interfaced to an Agilent 5973 mass spectrometer. The target compounds were separated on a DB- 5 MS capillary column (30 m x 0.25 mm I.D., 0.25 pm film thickness) with the following

Yang,Z., G. Zhang, B.P. Hollebone, C.E. Brown, C. Yang, P.G. Lambert, M. Landriault, and K. Shah, Fate of Oxygenated Components for Solar lnadiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar, Environment and Climate Chanoe Cenada. Ottawa. ON. oo. 415-440.2017.

076 419

oC oC oclmin temperature program: 60 for 2 min, heated to 300 at 6 and held for 20 min at 300oC. For all GC/MS analysis, ultrahigh purity helium was employed as carrier gas at a flow rate of 1 mllmin. The temperatures of injector, transfer line, ion source, and MS quadrupole oC, analyzer were held at290,290,300, and 150 respectively. Samples were injected in splitless/split mode. The mass-selective detector (MSD) was operated at an electron impact mode (70 eV) for selected ion monitoring (SIM) runs.

2.4.2. HPLC-HRMS Analysis of NAFCs Sample analysis was performed on Accela HPLC system paired with Exactive Orbitrap Mass System (Thermo Fisher Scientific, San Jose, CA, USA) with electrospray ionization (ESI) source in negative ion mode. A 20 min LC gradient method using a Poroshell l20EC-C¡ column (100 x 2.1mmi.d.,2.7 pm, Agilent) at a flow rute of 0.25 mllmin was used for separation. The gradient method was programmed among water with 0.0025% of formic acid (A); and 95%15% acetonitrilelwater with 3 mmol/L ammonium acetate (B) ,85%A0%15% isopropanol/toluene/water with 3 mmol/L ammonium acetate (C). In detail, 95Yomobile phase A and 5%o mobile phase C held isocratically for 2 min, followed by the linear gradients from 5olo to 100% mobile phase B in 8 min. Then mobile phase B decreased to l0% and C increased to 90/o linearly in the following 5 min, and held for 10 min. After that, mobile phase b increased to I00% in 0.5 min, held for 1.5 min, and then retumedto 5Yo B and 95% A in 0.5 min, and held for 5 min prior to the next injection. Orbitrap mass spectrometer was operated in ESI negative mode (ESI-) and data was acquired in full scan from the mass to charge (m/z) ratio of 80 to 1600. The negative ion of dimmer of acetic acid (2M-1, mlz:ll9.03498) was used as lock mass for scan-to-scan calibration correction to get less than 2 ppm mass accuracy. The mass parameters were set as follows: spray voltage, 4.0 kV; capillary temperature,32}oC; sheath gas,45; auxiliary gas, 15; and tube lens, -90 V. Heteroatom containing species including the formula of CflnN"Oo,S., where c, h, n, o and saretheelementalnumbersintheformula(0

DBE: +! +t (1) "-L22

Others were considered as isotope, adduct, minors and excluded in NAFC analysis. The reported NAFCs are the abundance of specific components normalized by the intensity of internal standard in each injection.

Yang,2., G. Zhang, B.P. Hollebone, C.E. Brown, C. Yang, P.G. Lambert, M. Landriault, and K. Shah, Fate of Oxygenated Components for Solar lrradiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar, Environment and Climate Chanoe Canada. Ottawa- ON. oo. 415-440.2017.

077 420

3. Results and Discussion 3.1. Variation of Carbonyl PAHs through Photo-oxidation Different carbonyl PAHs were formed after solar exposure in the two seasons, while they were non-detectable in all the control samples. For example, 1-indanone, 2-ethyl anthraquinone, benzo[a]fluorenone, 5,I2-naphthacenequinone, and fluorenones (C¡-FO, i from 0 to 4) were identified in most of the irradia-ted sa-mples. The qLra-ntified flr-rorenones, representing the carbonyl PAHs formed through photolysis, were plotted versus solar exposure time in the two seasons (Figure l). In summer exposure, fluorenones were formed in the beginning of solar exposure, while they were removed by the continual exposure. Different oil has different variation pattems due to their varied formation rates, which are dependent on the photo transformation rates of their precursors (e.g., PAHs and their alkylated congeners). Specifically, the maximal Ci-FO was observed after 5-d inadiation in ASMB#S, and 15-d in CLB and AV/B. More significantly decreasing trend was found in ASMB#S (after 5-d) and CLB than AWB (after 15-d), which suggests that more C,-FO were photo-oxidized in ASMB#S. The maximal abundance of the formed C¡FO was found in ASMB#S, followed by CLB and then AWB. It is consistent with the photol¡ic rates of those GC-detectable petroleum hydrocarbons reported by our former work (Yang et al., 2016), where ASMB#5 had the highest photolytic rates, followed by CLB and then AV/B. These findings indicate that these oxygenated PAHs were formed through the photo-oxidation of their parent PAHs; simultaneously, they were further photo- oxidized or mineralized. Once their formation rates were less than the photolytic rates, a decreasing trend was observed. Similarly, ASMB#5 has formed more G-FO in winter than CLB, while their variation trend is different from summer exposure depending on the specific oils. For example, fluorenones were formed in the first 30 days in ASMB#5 oil, while they were photolyzed by the continual summer exposure. They were formed continually during the 90 days solar exposure in the winter CLB. It is therefore concluded that oil property is one crucial factor affecting the variation of these oxygenated PAHs through photo-oxidation. It is reasonable because their precursors (e.g., PAHs or APAHs) showed different photolytic rates in different oils. Additionally, solar intensity and temperature also controlled the photolysis rates of PAHs and their alkylated congeners (Yang et al., 2016), which finally affected the formation and variation trend of these oxygenated PAHs in the two seasons. It is noted that only CLB, not AV/B was tested in the winter exposure, same to all the results addressed in the following sections. 3.2. Variation of NAFCs through Photo-oxidation The average negative-ion ESI Orbitrap mass spectra for the control and the 30 days summer irradiated NAFCs from the three test oils, ASMB#5, CLB and AWB, are depicted in Figure 2.It can be seen that the number of the identified peaks in the control sample of ASMB#5 is far less than the two dilbits (control), which suggests that more polar oxygenated acidic components are present in dilbits. It is reasonable because ASMB#S contains less resins and asphaltenes than the two dilbits (Yang et a1.,2016), while bitumen has suffered from extensive weathering process. The number of identified peaks increased significantly for ASMB#5, but not for the two dilbits, after 30-d summer inadiation. A relative shift in the most abundant peaks to lower mass defect and the increased absolute intensity were observed for all the irradiated samples, despite that the increased absolute intensity varied with the specific oil. For example, the most abundant peak in the irradiated ASMB#5 was about 5 times of its control, while only 2 times for CLB, and very slight enhancement for AWB. These observations indicate that the compositional

Yang, 2., G. Zhang, B.P. Hollebone, C.E. Brown, C. Yang, P.G. Lambert, M. Landriault, and K. Shah, Fate of Oxygenated Components for Solar lrradiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar, Environment and Climate Chanoe Canada. Ottawa. ON. oo. 415-440.2017.

078 421

complexity and the contents of NAFCs were enhanced after irradiation; simultaneously, abundant NAFCs with lower molecular weight have been formed after irradiation for all oils, especially for ASMB#5. Analyzing the peak numbers, ASMB#5 has formed more acidic components than the two dilbits because the abundant congeners are present in the original control dilbits. The more peaks and higher abundance of these acidic components in ASMB#5 suggest that its photolysis is more complete than the two dilbits, which is in agreement with our previous report (Yang et a1.,2016). We found that the target petroleum hydrocarbons, especially for aromatics, in ASMB#5, were degraded faster than those the two dilbits. The summer exposure also resulted in their faster degradation rates than winter for all oils. The physicochemical properties, solar intensity and temperature have been concluded to be the factors contributing to these differences. Unfortunately, low molecular weight NAFCs have been reported to be more susceptible to microbial activity or bioavailability (Frank et aL,2008; Frank et al., 2009).It is therefore that the final outcome will be the higher acute toxicity of the exposed aquatic system due to the presence of the abundant low molecular weight acidic components after solar irradiation.

3.3. The van Krevelen Diagrams for Compound Class Comparison after Solar Irradiation As mentioned in the experimental section, the average mass spectra combined with the assigned molecular formulas of NAFCs were generated in the present study. Based on the listed formulas, the van Krevelen diagram is one way to graphically display all compounds in a particular sample. It has been utilized to visualize complex organic samples by comparing the molar ratio of hydrogen to carbon (H/C) and oxygen to carbon (O/C) for major chemical compound classes, which can demonstrate the compositional differences between samples of different nature (coal versus petroleum), geographical origin (fossil fuels), processing (coal at two stages of liquefaction) (V/u et a1.,2004), and characterize biodegradation/photodegradation of crude oils (Islam et al., 2013; Ray et a1.,2014). Figures 3 and 4 compare the H/C and O/C ratios for the entire identified components with > 0.5o/o of the most abundant peak by HPLC-ESI (-)-Orbitrap MS before and after solar irradiation in the both seasons. It is well known that the H/C ratios can be used as the aromatic indices to visualize the differences of the unsaturated degree among samples (Koch and Dittmar,2005).In this diagram, the H/C differences (due to differences in DBE) are separated vertically. As the H/C ratio decrease, the number of rings plus double bonds increases. Same to the conclusions made by the previous sections, more components were formed after solar exposure. The components with larger O/C ratios but less H/C ratios were produced after irradiation in the both seasons. Specifically, most of the H/C ratios are ) 1.0 in the three control oils. However, a lot of components with H/C ratios ranging from 0.5 to 1.0 were formed after irradiation. Similarly, the heteroatom Oo classes (e.9., o from I to 8) are clearly separated horizontally by O/C ratio, which distinguishes classes differing in number of O atoms. In the present study, most of the O/C ratios are less than 0.4 in the three control samples. Some components with > 0.4 OIC ratios were formed after irradiation. It is obvious that solar irradiation has resulted in the oxidation of petroleum hydrocarbons by adding oxygen into them, suggesting abundant oxygenated components with higher water solubility and bioavailability were produced (Islam et al., 2013; Ray et a1.,2014). Therefore, abundant oxygenated components with higher number of rings or double bonds have been produced during the solar irradiation process.

Yang,Z., G. Zhang, B.P. Hollebone, C.E. Brown, C. Yang, P.G. Lambert, M. Landriault, and K. Shah, Fate of Oxygenated Components for Solar Inadiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar, Environment and Climate Chanoe Canada. Ottawa. ON. oo. 415-440.2017.

079 422

The comparison of winter and summer irradiation demonstrates that some of the newly formed components with less H/C ratios have been fuither decomposed after 60 to 90 days irradiation in summer, especially for ASMB#5. A generally increasing trend was observed for all winter inadiated samples. It is clear that these diagrams can let us qualitatively identify the difference among different samples, while the quantitative comparison needs further analysis.

3.4. Heteroatom Class Distribution through Photo-oxidation 3.4.1. Total Heteroatom Class Distribution as a Function of lrradiation Time In the present study, compounds were grouped into "classes" in a sample based on the number of heteroatoms by using assigned molecular formulas for each m/z peaks. Compositional trends among samples were visualized with heteroatom-class distribution graphs. Figure 5 shows the heteroatom class distribution for selected species with>lYo abundance of the internal standard in each injection in the ESI (-)-Orbitrap MS for the control and solar inadiated samples (from control to 90 days) in the two seasons. It is noted thationization suppression will cause an apparent reduction in anal¡e recovery from different matrices when ESI in negative ion mode is used (Page et a1.,2007). Unfortunately, no weathering-resistant biomarkers were available for Orbitrap MS to allow normalization across the data set in the present study. Herein, all the mass spectral peaks were noÍnalized to the abundance of the internal standard in each sample. This normalization can eliminate the variation of ionization efficiencies of compound classes across samples due to matrix effects. The most abundant acidic heteroatom class for all the control samples are components containing Oo species (i.e., compounds with the formula C"H¡O., where o ranges from 1 to 8). Specifically, AWB has the highest Oo species, followed by CLB, then ASMB#S. Similar trend is also suitable for the other sulfur- and nitrogen-containing species (SOo, S2Oo, NOo, NzOo, and SNnOo, where s and n span from I to 2). Significant amount of Oo species were formed after solar irradiation in the both seasons. The other species were formed too, while their hnal amounts were far less than the Oo species. Accordingly, the Oo species are the most abundant components in all the exposed samples. Among the three test oils, the formation of these acidic components is the highest in ASMB#5 in the two seasons, while the two dilbits have similar formation degree. Except for the most abundant Oo group in all the three oils, the N2O6 is the second abundant group, then NnSrOo group and the others in all the solar exposed ASMB#5 samples. In all the exposed CLB and AV/B, it seems that SOo is the second abundant group, followed by NOo group and then the others. It seems that the addition of oxygen has resulted in the formation of these oxygenated acidic components, while the formation of the sulphur or nitrogen containing acidic components is dependent on specific chemical composition of oil. In summer exposure, the total Oo species were continually formed until 30 days for all the three oils. The total Oo species in both ASMB#5 and CLB did not change significantly with the continual solar exposure until 90 days, similar to the SzOo, NOo and NzOo groups. The continual exposure has caused the decomposition of all groups in AWB and some of the groups in CLB (SOo and NOo) and ASMB#5 (SOo and NOo, and NnS.Oo). In winter, most of the groups were continually formed during the 90-day solar exposure. All the species after summer exposure are more abundant than those formed in winter. It is therefore concluded that oil property, exposure season (temperature and/or solar intensity) are the main factors affecting the variation of these heteroatom species during solar exposure.

Yang, 2., G. Zhang, B.P. Hollebone, C.E. Brown, C. Yang, P.G. Lambert, M. Landriault, and K. Shah, Fate of Oxygenated Components for Solar lrradiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar, Environment and Climate Chanoe Canada Ottawa ON oo 415-44O 2O17

080 423

3.4.2. Oo Acidic Class as a Function of lrradiation Time Oo class with different oxygen numbers (from 1 to 8) were subdivided from the total Oo species, despite that the alkyl-substituted aliphatic and cyclic monocarboxylic acids (O2 species), (Clemente and Fedorak,2005; Brown and Ulrich,2015), are the most abundant oxygen containing components in crude oil and diluted bitumen (Colati et a1.,2013). Higher-order oxygen classes (O+-Oz) have been reported from photo-oxidation (Ray et a1.,2014). Figure 6 compares the relative abundance for O1-Os acidic species in the control and inadiated samples. Without irradiation (control oils), the Oz class is the most abundant group; after irradiation, 02 class and the other higher-order oxygen classes (O:-OÐ increased in abundance significantly, while Or class did not see huge change. The variation of the different Oo species with irradiation time varied depending on oxygen number, specific oil and exposure season. Specifically, all the Oo species, except for Or, increased until 30-d irradiation in summer, a decreasing trend was observed with the continual irradiation for 02 species, while all the other higher-order O" (O3-Os) species either did not change significantly (e.g., all species in summer CLB), or slightly increased (e.g., Os, Oa, Ot and Os in summer ASMB#5) with the continual irradiation. Same to the observations mentioned in the previous section, summer exposure formed more abundant 02 to Os species than winter; ASMB#5 formed higher abundant Oz to Os species than CLB and AWB in the both seasons. All the Oo species have an overall increase during the winter irradiation, despite that their increase rates are far less than in summer. These findings further demonstrate that photochemical reactions increased the number of oxygen per molecular, thus increasing the water-solubility of oil-derived compounds, which is in agreement with previous study on Deepwater Horizon oil (Ray et al., 2014).

3.4.3. SrOo Acidic Class as a Function of Irradiation Time The distribution of S.Oo acidic class as a function of irradiation time in the two seasons is shown in Figure 7 . The SrOo acidic species are far less than the Oo species in both control and irradiated samples. In the control samples, SOo species are more abundant than SzOo species; SOz, SO:, and SO¿ species are the most abundant classes among the SOo species. The two dilbits have higher level of SrOo species than ASMB#S. The solar irradiation caused the formation of most of the SrOo species at the beginning of irradiation (the first 5 to 15 days exposure), such as, SO2, SO3, and SO¿ species, and SzO, SzOz, SzOs and SzO¿, while the continual exposure saw the decrease of these formed SrOo species in the two seasons. It seems that the varied exposure season did not affect the variation of these SrOo species. For different oils, the two dilbits have formed more abundant S.Oo species than ASMB#S. It is possible because dilbits have abundant sulfur-containing PAHs (e.g., dibenzothiophenes, and benzonaphthothiophenes) (Yang et al., 2016). The photochemical reactions of these parent sulfur-containing PAHs have resulted in the formation of more SrOo species (Bobinger and Andersson, 2009).

3.4.4. NnOo Acidic Class as a Function of lrradiation Time Figure 8 highlights the variation of NnOo acidic class with irradiation time. Without irradiation, ASMB#5 has higher level of NOo than NzOo species with NO, NOz and NOg as the most abundant classes, while the two dilbits have the most abundant NzO+ species, followed by N, NO, NOz and NO3 species. After irradiation, the relative distribution (normalized to 100% for all the NnOo species) of NnOo species shifted from left to right for all the three oils in the two seasons, which agrees with the increased oxygenation after irradiation. Specifically, more heavier-order NzOo species (e.g., N2O5-NzOs) and NOo species (from NOz to NO6) were formed for all of them. However, the variation patterns of NnOo species depends on oil properties and

081 424

exposure season. For example, the NO, NOz species showed a decreasing trend even after 5 days summer exposure in ASMB#5. However, the NOo species reached to the maximum after 5 days winter irradiation, then a decreasing trend was observed in the next 60 days for winter exposed ASMB#5. This discrepancy indicates that the formation and photolysis of these species occur simultaneously. In summer exposure, they were photo-oxidizedvery fast due to the stronger solar intensity and higher temperature compared to winter. It is therefore, that no increasing trend was observed after 5 days srünmer exposure for ASMB#5. Most of the other NOo species in the three oils increased at the first 5 to 15 days irradiation, and then decreased with the continual irradiation in the two seasons. The heavier-order NzOo species (1.{zO¿ to NzOs) were continually formed until 30 days in ASMB#S for the two seasons. However, the NzO+ species in the two dilbits decreased after inadiation, but the NzOs to NzOs species did not change after 5 days summer and 15 days winter irradiation. Based on the analysis of the variation of NnOo and S.Oo acidic classes, it can be concluded that the variation of these heteroatom-containing acids depends on the oil property, the contents of their parent compounds, and the exposed seasons.

3.5. Compositional Variation of Major Acidic Species through Photo-oxidation All the above discussed dataare based on the total or sub-total of the three major acidic species. As we have mentioned in the section 2,the elemental composition of these acidic species varies with the number of carbon, sulfur, oxygen and nitrogen, as well as the number of double bond/naphthenic rings. It is well known that each ring or double bond in a molecule reduces its number of hydrogen atoms by two (Lemkau et al., 2014). DBE, representing the number of rings plus double bond, is a very useful parameter to examine the compositional differences across samples. Therefore, the variation of the elemental composition of major acidic species derived from Orbitrap-MS characterization through photo-oxidation can be compared by color-coded isoabundance contour plots of DBE versus carbon number. Figure 9 shows the isoabundance contoured plots of DBE versus carbon number for the most abundant acidic classes (Oo species from O to Os) in the three test oils, which highlights the compositional changes that occurred over the 90 days solar exposure and used to compare the compositional changes within a heteroatom class. By analyzing the color codes, it is clear that all the solar irradiated samples were found to have higher level of Oo species than their corresponding controls; the color spectrum became wider in both vertical and horizontal directions, suggesting that higher abundance and more acidic Oo species were formed after solar irradiation. For different seasons, summer exposure formed higher abundance and more acidic species than winter. An overall increasing abundance was observed for the winter CLB and ASMB#5, while they increased at the beginning, and then decreased/unaltered with the continual inadiation for all summer inadiated samples. Among different oils, the two dilbits have more acidic Oo species and higher abundances than ASMB#5 in the control samples due to their long term weathering process. ASMB#5 has faster formation rates and higher abundances than the two dilbits in the two seasons, which is also in agreement with the photolysis rates of their precursors (Yang et al., 2016). The shape of color plots is flatter vertically in CLB and AWB than ASMB#5 in all the exposed samples. It seems that the solar irradiated oils (especially for ASMB#5) have formed abundant NAFCs with > 4 DBEs at the beginning of solar inadiation (30 days in winter and 15 days in summer), while some of them lost their abundance with the continual irradiation. These phenomena further confirm that the formation and photolysis of these components occurred simultaneously during inadiation. As reported in our former report (Yang et a1.,2016), the precursors in summer ASMB#S have been exhaustively depleted after 15 to 30 days irradiation. The photolysis loss of

082 425

these photo-sensitive Oo species could not be reimbursed by the photolysis of their precursors in this scenario, which finally depicted à decreased trend. The maximal abundant Oo species in all the inadiated samples shifted to left by horizontal direction and down by vertical direction. This suggests that the newly formed Oo species are predominantly composed of less double bonds or naphthenic rings and less carbon numbers. These findings are in agreement with some previous reports for water-soluble acidic extracts detected after solar irradiation of crude oil (Ray et al., 2014), and the fate of NAs with monocarboxylic acids (Oz-NAs) in one of our work (Yang et al., to be published). It was found that the Oz-NAs in the solar irradiated samples were abundant with the congeners constituting lighter molecular weight and less cyclic rings; the maximal O2-NAs in the inadiated samples shifted to smaller carbon number. It seems that photo degradation of NAs or NAFCs preferably photolyzes the components with more cyclic rings due to their photo-sensitivity. This phenomenon is opposite to the biodegradation of NAFCs, the congeners with more cyclic rings were reported to be the most environmental persistent and resistant to biodegradation (Misiti et a1.,2014). The color-coded isoabundance contour plots of DBE versus carbon number can identify some variation trends; however, it is difficult to reveal the variation of the study targets with carbon and DBE number. The 2-dimensional plots between the relative intensity and DBE/carbon number will be discussed in the following sections to address the distribution profiles of Oo species along carbon and DBE number through the whole photo-oxidation process.

3.6. Variation of Total Acidic Oo Species versus Carbon Number through Photo- oxidation The plots of the relative abundance of total acidic Oo species versus carbon number (from Ca,to Ceo) are shown in Figure 10. A typical bell shape with the maximal species in the carbon range of 16 to 24 in all the oils was observed. Similar to the above findings, solar irradiation formed abundant acidic Oo species in specific carbon range for different oils. Specifically, abundant Oo species were formed in the carbon range of less than C5s, Cqo, ãnd Czs inthe suÍrmer ASMB#5, CLB and AV/B, respectively. In winter, abundant components were found for carbon number < 50, and 20 in ASMB#5 and CLB after irradiation. No significant change was observed for the congeners with > Cso ttrough inadiation in all the oils. The discrepancy for same oil exposed in the two seasons can partially attribute to the differences of solar intensity and temperature. The maximal abundant Oo species shifted from right to left after solar inadiation (e.g., from Cz¿, to Cre for A\MB, Czt to Crc for CLB, and Czo to Cs-Crc for ASMB#S), which suggests more acidic Oo species with light molecular weight were formed through photo- oxidation.

3.7. Variation of Total Acidic Oo Species versus DBE through Photo-oxidation The plots of the relative abundance of total acidic Oo species versus DBE (from I to 16) are shown in Figure 1 1. Along the DBE values, the abundance of all the Oo species decreased with the increase of DBE values or cyclic rings in the control ASMB#5. However, the Oo species with DBE of 3 (2 cyclic rings) is the most abundant group in the control AWB and CLB, followed by DBE:2 (one cyclic ring), and 1 (acyclic ring), then shown an overall decreasing trend with the increase of DBE. After irradiation, the most abundant group is the one with a DBE of 2 (i.e.,1 double bond or one cyclic ring), followed by acyclic group, DBE:3, and 4, a hump from DBE:4 to I (3 to 7 cyclic rings or double bonds), and then an overall decreasing trend with the increase of DEB values for all the other DBEs in most of the exposed oils. It is clear that,

083 426

except for the abundant Oo species with one cyclic or non-acyclic ring, abundant congeners with 3 to 7 cyclic rings were also formed during solar irradiation. The components constituting 3 to 7 cyclic rings may be derived from the photolysis of some hydrocarbons with polycyclic aromatic rings because these precursors are the most photo-sensitive groups (Yang et a1.,2016). The general distribution patterns of the Oo species are similar to those of Oz-NAs reported in another research (Yang et al., to be published), except that the Ø-NAs without double bond or cyclic ring are the most abundant species among all the Oz-NAs. This discrepancy can be ascribed to the target difference. As discussed in previous section, abundant Oo species with higher number of oxygen (3 to 8) were formed through solar irradiation. The contribution from the O3to Os NAFCs may slightly change the distribution patterns of the total Oo species. A significantly decreased trend was observed for the Oo species with DBE:5 to 8 in the summer ASMB#5 and AWB after 60 and 90-day irradiation, but not for the suÍrmer CLB. Congeners with lower DBE values (1 and 2) increased at the beginning of inadiation in summer, then they almost un-altered with the continual irradiation. It seems that the photo-stability of Oo species depends on the DBE values. Generally, the higher DBE is, the less photo-stability of Oo species is. The decrease of the congensrs with higher DBE values in summer ASMB#5 and AWB can be partially ascribed to the following possibilities. Firstly, ASMB#5 was photolyzed fastest among the three oils in summer (Yang etal.,20l6), the depletive consumption of the photo-sensitive precursors could not reimburse the photolytic loss of these formed acidic components. Secondly, the photolysis rates of the petroleum hydrocarbons in AWB are the lowest among the three oils, it is therefore that the formation rates of these acidic components are limited compared to their photolysis rates. As mentioned earlier, acidic extracted organic species with more cyclic rings are the most environmental persistent and resistant to biodegradation (Misiti etal.,2014). The preferable photolysis of the acidic species with more cyclic rings may supply an alternative remediation technology for spilled oil residues. 4. Conclusions The fate of the identified oxygenated PAHs and polar acidic extractable fractions with the elemental composition of C.H¡O.SN' was investigated by exposing diluted bitumen to the natural sunlight in saltwater. Various carbonyl PAHs were identified, while fluorenones were the most abundant components identified. Abundant polar acidic components with the most abundant species of Oo (from Oz to Os) were formed after irradiation; same to the SrOo and NnOo species, despite their level was far less than the Oo class. Abundant heavier-order SrOo and NnOo species were formed after irradiation compared to their controls, especially for some NzOo and SOo species. The formed acidic Oo components shifted to less carbon numbers, less DBE values, and higher oxygen numbers. The variation of these oxygenated components with the solar inadiation time depends on the oil property and exposure season. Generally, ASMB#5 formed higher acidic components than the two dilbits, same to the summer irradiation, which is in agreement with the photo-oxidation trends of some GC-detectable petroleum hydrocarbons. Except for studying the GC-detectable petroleum hydrocarbons, this study provides an altemative way to investigate the fate of dilbits, and supplies chemical evidence for toxicity assessment of aquatic environment after dilbit products spill.

5. Acknowledgements This work was funded and supported by the Government of Canada's World Class Tanker Safety System (WCTSS) program and Ocean Protection Plan program. Great thanks are also expressed to Oni Olatunji and Suhash Aravindan for editing programs for data analysis.

Yang,Z., G. Zhang, B.P. Hollebone, C.E. Brown, C. Yang, P.G. Lambert, M. Landriault, and K. Shah, Fate of Orygenated Components for Solar lrradiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar, Environment and Climate Chanoe Canada. Ottawa. ON. oo. 415-440.2017.

084 427

6. References Aeppli, C., C.A. Carmichael, R.K. Nelson, K.L. Lemkau, Graham W.M., M.C. Redmond, Valentine D. L. and R.C. M., "Oil Weathering after the Deepwater Horizon Disaster Led to the Formation of Oxygenated Residues", Environmental Science & Technologt, 46:8799-8807, 20t2.

Bobinger, S. and J.T. Andersson, "Photooxidation Products of Polycyclic Aromatic Compounds Containing Sulfur", Environmental Science & Technologl,t, 43:8119-8125,2009.

Brown, L.D. and A.C. Ulrich, "Oil Sands Naphthenic Acids: A Review of Properties, Measurement, and Treatment", Chemospher e, 127 :27 6-290, 20I 5.

Clemente, J.S. and P.M. Fedorak, "A Review of the Occurrence, Analyses, Toxicity and Biodegradation of Naphthenic Acids ", Chemo sphere, 60:585 -600, 2005.

Colati, K.A.P., G.P. Dalmaschio, E.V.R. de Castro, A.O. Gomes, B.G. Yazand'W. RomÃfo, "Monitoring the Liquid/Liquid Extraction of Naphthenic Acids in Brazilian Crude Oil Using Electrospray Ionization Ft-Icr Mass Spectrometry (ESI FT-ICR MS)", Fuel, 108:647-655,2013

Crosby, S., R. Fay, C. Groark, A. Kani, J.R. Smith, T. Sulivan and R. Pavia, "Transporting Alberta Oil Sands Products: Defining the Issues and Assessing the Risks". US Department of Commerce (Ed.), NOAA technical memorandum, Seattle, WA, 2013.

D'Auria, M., L. Emanuele, R. Racioppi and V. Yelluzzi, "Photochemical Degradation of Crude Oil: Comparison between Direct Irradiation, Photocatalysis, and Photocatalysis on Zeolite", Journal of Hazardous Material s, I 64:32-38, 2009.

D'Auria, M., R.Racioppi and V. Yelluzzi, "Photodegradation of Crude Oil: Liquid Injection and Headspace Solid-Phase Microextraction for Crude Oil Analysis by Gas Chromatography with Mass Spectrometer Detector", Journql of Chromatographic Science, 46:339-344, 2008.

Frank, R.4., K. Fischer, R. Kavanagh and K.R. Solomon, "Effect of Carboxylic Acid Content on the Acute Toxicity of Oil Sands Naphthenic Acids", Environmental Science & Technologt, 43:266-271,2009.

Frank, R.4., R. Kavanagh, B.K. Bumison, G. Aresnault, J.V. Headley, K.¡4. Peru, G. Van Der Kraak and K.R. Solomon, "Toxicity Assessment of Collected Fractions from an Extracted Naphthenic Acid Mixture", Chemosphere, 72:1309 -13 | 4, 2008.

Halton, K., A.S. Catherine, D.G. Burrows, T.K. Collier, N.L. Scholz, and J.P. Incardona, "Natural Sunlight and Residual Fuel Oil Are an Acutely Lethal Combination for Fish Embryos", Aquatic Toxicologt, 99 :56-64, 2010.

Headley, J.V., M.P. Barrow, K.M. Peru, B. Fahlman, R.A. Frank, G. Bickerton, M.E. McMaster, J.L. Parrott and L.M. Hewitt, "Preliminary Fingerprinting of Athabasca Oil Sands Polar Organics in Environmental Samples Using Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry", Rapid Communication Mass Spectrometry,25:1899-1909, 2011.

085 428

Headley, J.V., K. Peru, B. Fahlman and D.W. McMartin, "Selective Solvent Extraction and Characterization of the Acid Extractable Fraction of Athabasca Oil Sands Process Waters by Orbitrap Mass Spectrometry", International Journal of Mass Spectrometry,345-347:104-108, 2012.

Hughey, C.4., S.A. Galasso and J.E. Zumberge, "Detailed Compositional Comparison of Acidic Nso Compouncis in Bio

Incardona, J.P., T.L. Swarts, R.C. Edmunds, T.L. Linbo, A. Aquilina-Beck, C.A. Sloan, L.D. Gardner, B.A. Block and N.L. Scholz, "Exxon Yaldezto Deepwater Horizon: Comparable Toxicity of Both Crude Oils to Fish Early Life Stages", Aquatic Toxicology, 142-143:303-316, 2013.

Islam,4., Y.Cho, U.H. Yim, W.J. Shim, Y.H. Kim and S. Kim, "The Comparison of Naturally Weathered Oil and Artificially Photo-Degraded Oil at the Molecular Level by a Combination of Sara Fractionation and FT-ICR MS", Journal of Hazardous Materials, 263,Part2:404-4l1, 20t3.

John, G.F., Y. Han and T.P. Clement, "Weathering Patterns of Polycyclic Aromatic Hydrocarbons Contained in Submerged Deepwater Horizon Oil Spill Residues When Re- Exposed to Sunlight", Science of The Total Environmenf 573:189-202,2016.

Koch, B.P. and T. Dittmar, "From Mass to Structure: An Aromaticity Index for High-Resolution Mass Data of Natural Organic Matter", Rapid Commanication Mass Spectrometry,20:926-932, 2005.

Lemkau, K.L., A. McKenna, D.C. Podgorski, R.P. Rodgers and C.M. Reddy, "Molecular Evidence of Heavy-Oil Weathering Following the M/V Cosco Busan Spill: Insights from Fourier Transform Ion Cyclotron Resonance Mass Spectrometry", Environmental Science & Technology, 48:3760-3767,2014.

Maki, H., T. Sasaki and S. Harayarø, "Photo-Oxidation of Biodegraded Crude Oil and Toxicity of the Photo-Oxidized Products", Chemosphere, 44:1145-1151, 2001.

Mallakin,4., B.J. McConkey, G. Miao, B. McKibben, V. Snieckus, D.G. Dixon and B.M. Greenberg, "Impact of Structural Photomodification on the Toxicity of Environmental Contaminants: Anthracene Photooxidation Products", Ecotoxicologt and Environmental Safety, 43:204-212,1999.

Mclafferty, F.W. and F. Turecek, "Interpretation of Mass Spectra, 4th Ed." University Science Books, Mill Vallet, CA, 1993.

Misiti, T.M., U. Tezel and S.G. Pavlostathis, "Effect of Alkyl Side Chain Location and Cyclicity on the Aerobic Biotransformation of Naphthenic Acids", Environmental Science & Technologt, 45:7431-7437,2014.

086 429

Page, J.S., R.T. Kelly, K. Tang and R.D. Smith, "Ionization and Transmission Efficiency in an Electrospray Ionization-Mass Spectrometry Interface", Journal of the American societyfor Mass Spectrometry, 1 8: 1 582- 1 590, 2007 .

Ray,P.Z., H. Chen, D.C. Podgorski, A.M. McKenna and M.A. Tarr, "Sunlight Creates Oxygenated Species in Water-Soluble Fractions of Deepwater Horizon Oil", Journal of Hqzardous Materials, 280:636-643, 201 4.

Rowland, S.J., A.G. Scarlett, D. Jones, C.E. West, R.A. Frank and L.M. Hewitt, "Sterodial Aromatic'Naphthenic Acids'in Oil Sands Process-Affected Water: Structural Comparisons with Environmental Estrog eîs", Environmental Science & Te chnolo 94 45 : 9806-98 I 5, 201 1 .

Stanford, L.A., S. Kim, G. Klein, D.F. Smith, R.P. Rodgers and A.G. Marshall, "Identification of V/ater-Soluble Heavy Crude Oil Organic-Acids, Bases, and Neutrals by Electrospray Ionization and Field Desorption Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry", Environmental Science & Technolo g¿ 4l :2696-27 02, 2007 .

Vaughan, P.P., T. Vy'ilson, R. Kamerman, M.E. Hagy, A. McKenna, H. Chen and V/.H. Jeffrey, "Photochemical Changes in Vy'ater Accommodated Fractions of MC252 and Sunogate Oil Created During Solar Exposure as Determined by FT-ICR ly'rS", Marine Pollution Bulletin, 104:262-268, 2016.

Wang, Z'C.Yang,Z.Yang, C.E. Brown, B.P. Hollebone and S.A. Stout, "Petroleum Biomarker Fingerprinting for Oil Spill Charactenzation and Source ldentification", In: Stout, S.4., Wang, Z. (Eds.). Standard Handbook Oil Spill Environmental Forensics-Fingerprinting and Source Identification. Elsevier Inc., London, UK, pp. 252-270,2016.

Wu,Z., R.P. Rodgers and A.G. Marshall, "Two- and Three-Dimensional Van Krevelen Diagrams: A Graphical Analysis Complementary to the Kendrick Mass Plot for Sorting Elemental Compositions of Complex Organic Mixtures Based on Ultrahigh-Resolution Broadband Fourier Transform Ion Cyclotron Resonance Mass Measurements", Analytical Chemis try, 7 6 :25 | I -25 | 6, 2004.

Yang, Z.,B.P. Hollebone, C.E. Brown, C. Yang, Z.Wang,G.Zhang,P. Lambert, M. Landriault and K. Shah, "The Photolytic Behavior of Diluted Bitumen in Simulated Seawater by Exposed to the Natural Sunlight", Fuel, 186:128-139, 201 6.

Yang, Z.,B.P. Hollebone, G.Zhang, C.E. Brown, C. Yang, P. Lambert,Z.Wang, M. Landriault and K. Shah, "Fate of O2-Naphthenic Acids for Photodegraded Diluted Bitumen in Seawater", to be published.

087 430

120 --a- 100 CLB --+* Cr-Fluorenone 80 .r Cr-Fluorenone 60 --r* C.-Fluorenone 40 Co-Fluorenone 20 --'õ0 .C o) 20 40 60 80 '100 250 3è) c zoo ASMB#5 .o 150 1oo Ec E50 o5o õ o 0 20 40 60 80 100 E60 so AWB 'î 40

30 t ---- 20 10 0

0 20 40 60 80 100 (b) 120 100 CLB

t.--- -.. . =Boo .c 60 940 920 ;oo (E 0 20 40 60 80 100 Þ 250 C zoo ASMB#5 c$ o 150 õ 100 o) Eso 'îo 0 20 40 60 80 100 Solar Exposure time (days)

Figure 1 Variation of carbonyl PAHs through photo-oxidation. (a)n summer; (b), winter.

088 431

ASMB#5 400t Control 3{n0

2fl0ü ü) (J c 100ù fit E 0 c Ðmo Ë:l Sfl { 1500{¡ 1û{Xl0

5{XX¡

{t äl{l 4{10 ffi0 S0û lü¡fl0 CLB 4{tgo Cor¡lrol 3{n0 0t ãn{t (J E n 1Ð00 E C 0 3 sÐûo le 30d Ëü¡t düto

zfm0

0 ät0 .fl0 Éffi 80t 1000 AWB 6û00 Sfmt Control {g!ü {¡ ädnI (J f ilno n¡ E 1{r00 E t f g{Ðü _û 30d frÐ0 4SO

2{¡00

{t âlo 40{t ffit 8{¡0 1üfl! mf¿

Figure 2 Average negative-ion ESI Orbitrap mass spectra of the acidic extractable fractions in the control and the 30 days irradiated in summer from the three test oils, ASMB#S, CLB and AWB.

089 432

CLB AWB ASMB#5

5d-control Total (>0.5%)

3.0 3.0 3.0

2.5 2.5 2.5 a aaa .o a aa 2.0 2.0 a 2.0 aa a g a ao a a 1.5 tP "t.5 1.5 a C) I 1.0 1.0 a 1.0 0.5 0.5 a 0.5 a a ¡ a a 0.0 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

30d-total (> 0.5%)

3.0 3.0 3.0 t a aa 2.5 2.5 a 2.5 aa o 2.0 2.0 2.0 a ñ a L 1.5 1.5 1.5 O I 1.0 1.0 1.0 0.5 0.5 a 0.5 a 0.0 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

60d-Total (>0.5%)

3.0 3.0 3.0 2.5 t 2.5 2.5 o I 2.0 2.0 2.0 Lo o 1.5 1.5 1.5 I 1.0 1.0 1.0 0.5 0,5 0.5 0.0 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

90d-Total (>0.5%)

3.0 3.0 3.0 a a a 3 2.5 I 2.5 a 2.5 o - 2.0 2.0 2.0 Lo o 1.5 1.5 1.5 T 1.0 1.0 1.0 a 0.5 0.5 0.5 0.0 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8'1.0 O/C ratio O/C ratio O/C ratio

Figure 3 Van Krevelen plots for the total ions of the acid fraction analyzed by O ESI- Orbitrap-MS in the representative summer control and exposed samples

090 433

CLB ASMB#5

5d-control Total (>0.5%)

3.0 3.0 2 .5 2.5 aa o a a 2,0 a t 2.0 a aa G I L aa a t ta o 1 .5 a 1.5 a I ,| .0 ll 1.0 lo a a a a a a a ¡l 0.5 0.5 a a I t a a 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

3Od-total(> 0.5%)

3.0 3.0 a at I 2.5 a fa 2.5 a 2.0 2.0 €(5 ll¡ L a (J 1.5 1.5 ì 1.0 1.0 0.5 a aa 0.5 a 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

60d-Total (>0.5%)

3.0 3.0 a I 2.5 t 2.5 2.0 2.0 €(5 L 1.5 a 1.5 O a ì 1.0 1.0 0.5 a 0.5 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

9Od-Total (>0.5%)

3.0 3.0 a 2 .5 2.5 ra o a 2 .0 2.0 aa E a 1 .5 1.5 o ,| 3¡ I .0 1.0 0.5 0.5 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 O/C ratio O/C ratio

Figure 4 Van Krevelen plots for the total ions of the acid fraction anaþed by (-) ESI- Orbitrap-MS in the representative winter control and exposed samples

Yang,Z, G. Zhang, B.P. Hollebone, C.E. Brown, C. Yang, P.G. Lambert, M. Landriault, and K. Shah, Fate of Oxygenated Components for Solar lrradiated Diluted Bitumen in Saltwater, Proceedings of the Fortieth AMOP Technical Seminar, Environment and Climate Chanoe Canada. Ottawa. ON. oo. 415-440. 20'17.

091 434

(a) 500 I 400 I 5d 300 ASMB#5 t 15 d 200 rlll I 30d I 60d I 90d

0 Oo OoS OoS2 OoN OoN2 OoSsNn ¡ 300 '-(t 250 c CLB o 200 .g 150 I 1oo a5 50 õ ú.u Oo OoS OoS2 OoN OoN2 OoSsNn 300 250 AWB 200 150 100 50 0 Oo OoS OoS2 OoN OoN2 OoSsNn (b) 140 120 100 ASMB#5 80 60 à40g. Po20 .ç Oo OoS OoS2 OoN OoN2 OoSsNn o iu 120 P'38 CLB 60 40 20 0 Oo OoS OoS2 OoN OoN2 OoSsNn Representative species

Figure 5 Variation of the major acidic classes observed of the control and solar exposed samples in the both seasons. (a), summer; (b), winter.

092 435

(a) 200 Control 150 5d ASMB#s 15 d 100 30d 60d 90d 50

0 oo203o405060708 120

=U' c CLB o 80 .c o 40

(õ o 0 É. oo203o,405060708 150

AWB 100

0 oo.203o'405060708

(b) 50 40 ASMB#5 30 'õ c 20 .910c õ0 oo203o,405060708 Ë60 Puo CLB 40 30 20 10 0 oo203o'405060708 Oo species

Figure 6 Variation of acidic Oo series observed of the control and solar exposed samples in the both seasons. (a), summer; (b), winter.

093 436

(a) SO Sroo o 4 2.0 ASMB#5 I CNT ASMB#5 1.6 3 I 5d 15d ffin 1.2 2 I 30d I 60d .8 I 90d 1 .4 0 0.0 S 63 s2S s3S s+S s59 s6S o'lS oBS SZ gS2 s2SZ 63S264526552665261 S26s32 25 3.0 CLB CLB =tt 20 2.5 c 2.0 c) 15 ._c 1.5 10 o 1.0 5 (tr .5 õ 0 0.0 É. S gS 625 g3S 645 g5S 965 g1S gss 52 9S26252935294329g32905291S26952 25 2.0 AWB AWB 20 1.6

15 1.2

10 .8

5 .4

0 0.0 S 93 52 623 6g3 94S o5S o65 o1S oBS (b) 6S262S2s3S2saS295S29oS2gt92gsS2

2.0 1.5 ASMB#5 ASMB#5 1.5 1,0 '1.0

5 =U' c o .= 0.0 0.0 S 95 6!5 g3S 645 653 605 615 6eS 32 9329232993294529552905291 S29OS2 o 25 2.5 CLB CLB (5 õ 20 2.0 É. 5 1.5

0 1.0

5 .5

0 0.0 S 65 92S 939 g4S g5S 665 615 6eS 32 6329252635264529532965291529952

S"Oo species

Figure 7 Variation of acidic SrOo series observed of the control and solar exposed samples in the both seasons. (a), summer; (b), winter.

094 437

(a) Noo Nroo 2.5 I 14 ASMB#5 12 ASMB#5 2.0 I 5d se$ 15d 10 1.5 I 30d I I 60d 6 1.0 I 90d 4 .5 2 0.0 0 N 6N 92N 93N gnN 95N s6N slN gsN N2 /, r-(J¿¡--(JJrr-(J*r -rucr--ljo¡ --1J I \ --( øt¿ 12 CLB 6ro '10 .cg I 96 6 õ4 4 d)t 2 É. Ïr fl,fl il ¡ 0 0 N oN oZN qlN 6aN o6N oON olN ogN N2 9N292N2¡3N2s+Nfo sN26oNþ N2ssNZ 10 25 AWB I 20 6 15

4 10

2 5

0 0 NoN 63N 64N 65N o?N oBN N2 eNfo 2N2e3N26N26oNfo oNþ N26sN2 (b) 6 3.5 ASMB#5 ASMB#5 5 3.0 4 2.5 2.0 3 1.5 Ë 2 1.0 a ,| c .5 o 0 0.0 .c N 9N q2N q3N sAN 95N soN slN gsN ¡12 6N2e2N2erNfo aNfo sNfooNþ o 16 10 14 CLB (5 12 I õ 10 6 É. I 6 4 4 2 2 0 0 NoN 63N 6AN 65N S2 oN2ozNzo¡NzoaNþsNfooNþNzeoNz

NnOo species

Figure I Variation of acidic NnOo series observed of the control and solar exposed samples in the both seasons. (a), summer; (b), winter.

095 438

WASMB#s WCLB SASMB#s SCLB SAWB 301 . - tol- r-- 20 ---- ;:---- s o-crur fi 'o -30 ','o ; -.n¡r ^t & î|,e+'l 20 40 60 20 40 60 20 40 60 20 40 60 20 40 60 30 30

uJ 20 r 20 5d dl tt o ----.l 10

0 þf¡b i 20 40 60 30 ,30.. --'.. .. _

l.o i.o I E20 -l I zo- iro. ''t5 d ul0F '* i 0 t+--+ 40 60 20 40 60 20 40 60 40

tJJ 30d g

20 40 20 40 20 40 60 40 60 30 30

20 20 ï---- ] o.o2 I.JJ 20 60 d co o 10 10

0 0 0 20 40 60 20 40 60 60 *rÞr 30 30

tll 20 eod 3 '10 10 -

0 0 20 40 60 20 40 Carbon number Carbon number Carbon number Carbon number Carton number

Figure 9 Isoabundance-contour plots of double band equivalent (DBE) vs. carbon number for acidic Oo species derived from negative-ion ESl-Orbitrap-mass spectra in the control and solar irradiated samples in winter (left two) and summer (right three). Notes: WASMB#5 means winter exposed ASMB#5, SASMB#5 means summer exposed ASMB#5, same to CLB and AWB; CNT means control sample.

096 439

(a)

30 25 Control ASMB#5 5d 20 '15 d 15 30d 60d 10 90d 5 0

6 =U' c 4 (¡) 2 CLB .c 0 o I F 6 ¡ì -g 4 o 2 É. 0 18 l6 14 AWB 12 10 I 6 4 2 0

(b) 10 I 5d I ASMB#5 15d 6 30d 60d 4 90d à2a Eoc õ10

ÈB(5 CLB õ6 É. 4

10 20 30 40 50 60 70 Carbon number

Figure 10 Variation of total acidic Oo species versus carbon number observed of the control and solar irradiated samples in the both seasons. (a), summer; (b), winter.

097 440

(a) 120 100 .1..t + Control ú ASMB#S . 5d 80 ^'\' \r 15 d -, 'ù\ \\... 4- - 30d 60 - .,!. 60d 40 . t-'*å -+- . * --F- 90d 20 ' ; '-ìl .:. 0 :+*-.¡_¡È -{ .à 60 so ?. CLB 'õË40 l^t 30 \\ "l\ R È20 t-\ -Ë 10 -t.- ìÈ É.0

60 .^ 50 ¿- \ AWB 40 r\{. 30 ì * ìl 20 -.\ 10 0

(b) 25 \ + Control 20 \ ASMB#5 5d 15 d 15 +- - - 30d \\ \ 60d àto - -+- o l\-'l/ -.-ts - 90d c5 .:=|, . 0) Eo *'-.1-.-r---.¡ o

ts * ã20 'lt É' CLB 15 \ r- -)tts \ 10 ì. 5

0246810 12 14 16 18 DBE

Figure 11 Variation of total acidic Oo species versus DBE observed of the control and solar irradiated samples in the both seasons. (a), summer; (b), winter.

098 Environmental Pollution 231 (2017) 622e634

Contents lists available at ScienceDirect

Environmental Pollution

journal homepage: www.elsevier.com/locate/envpol

Fate of oxygenated intermediates in solar irradiated diluted bitumen mixed with saltwater*

* Zeyu Yang , Gong Zhang, Bruce P. Hollebone, Carl E. Brown, Chun Yang, Patrick Lambert, Zhendi Wang, Mike Landriault, Keval Shah

Emergencies Science and Technology Section, Science and Technology Branch, Environment and Climate Change Canada, Ottawa, ON, K1A0H3, Canada article info abstract

Article history: Two types of diluted bitumen (dilbit) and a light crude oil spiked onto the surface of saltwater were Received 17 May 2017 irradiated with natural solar light in Ottawa to assess the impact of sunlight to the fate of oxygenated Received in revised form intermediates. Oxygenated components, including carbonyl polycyclic aromatic hydrocarbons (PAHs) 4 August 2017 and acidic polar fractions (naphthenic acid fraction compounds, NAFCs), were identiﬁed after periods of Accepted 11 August 2017 solar exposure under both winter and summer conditions. Carbonyl PAHs and NAFCs were formed in Available online 29 August 2017 both seasons; however, light crude and summer irradiation produced higher abundance of them than dilbits and winter exposure. The formed NAFCs were abundant with the congeners containing a het- Keywords: Carbonyl PAHs eroatom of oxygen only (Oo species), accompanied by the minor amounts of sulfur- and nitrogen- Naphthenic acid fraction compounds containing acids. The produced Oo species were predominant with the congeners with light molecular High resolution LC-MS weight, high degree of saturation and heavy oxygen numbers. For both carbonyl PAHs and NAFCs, their Diluted bitumen abundance continually increased throughout the period of winter exposure. In the summer, some Solar irradiation carbonyl PAHs and all Oo species increased during the early exposure period; then they decreased with continued exposure for most oils, illustrating their transitional nature. Oxygenated intermediates thus appear to have been created through the photo-oxidation of non-to medium-polar petroleum hydrocarbons or the intermediates of aldehydes or ketones (O1). Oil properties, the duration of exposure, exposure season and the chemical structure of these intermediates are critical factors controlling their fate through photo-oxidation. The observed chemical changes highlight the effects of sunlight on the potential behavior, fate and impact of spilled oil, with the creation of new resin group compounds and the reduction of aromatics and saturates. These results also imply that the ecological effects of spilled oil, after ageing in sunlight, depend on the speciﬁc oil involved and the environmental conditions. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction from sunlight represents an important pathway inﬂuencing the transformation and fate of oil components in the environment. Diluted bitumen (dilbit) is an oil sand product produced by the Oil ﬁngerprinting technology uses gas chromatography (GC) or mixture of 20e30% light condensates in bitumen (Crosby et al., gas chromatography-mass spectrometry (GC-MS) to characterize 2013). Its environmental fate and behavior has drawn extensive the components of petroleum hydrocarbons and evaluate the ratios attention from environmental chemists. As with conventional of selected ‘diagnostic’ compounds. This technology has been used crude oils, once spilled into the marine environment, dilbit extensively to investigate the photolytic fate of GC-detectable pe- immediately begins to undergo a variety of physical, chemical and troleum hydrocarbons resulting from the irradiation of conven- biological processes, such as evaporation, dispersion, photo- tional crude oil or diluted bitumen (D'Auria et al., 2008; Radovic oxidation and microbial degradation. Among these processes, et al., 2014; Wang et al., 2016; Yang et al., 2016). It has been photochemical degradation caused by the adsorption of photons shown that abiotic and biotic weathering processes produce oxygenated transformation products that account for 60e80% of extractable material (Aeppli et al., 2012). Photo-oxidation most

* This paper has been recommended for acceptance by Charles Wong. likely enhances the dispersion of oil spills by transforming the oil * Corresponding author. into more water-soluble oxygenated components (King et al., 2014; E-mail address: [email protected] (Z. Yang). http://dx.doi.org/10.1016/j.envpol.2017.08.043 0269-7491/© 2017 Elsevier Ltd. All rights reserved.

099 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634 623

Ray et al., 2014; Wang et al., 2016). A very large discrepancy exists Canada, 45.4N) over the winter and summer seasons. AWB was not between the abundance of the conventional GC-detected petro- tested over winter as it has similar physicochemical properties as leum hydrocarbons (e.g., PAHs) and the associated biological CLB. GC and GC/MS analysis was used to assess the variation of TPH, response. This suggested that unidentified compoundsdand not alkanes, aliphatic petroleum biomarkers, and APAHs through only PAHsdare responsible for the adverse biological responses photo-oxidation. We showed that the chemical structure of pe- (Halton et al., 2010; Incardona et al., 2013). Studies have concluded troleum hydrocarbons controlled the photolysis of different that after photo-oxidation, oil has more adverse effects on algae, chemical groups in the dilbits and ASMB (Yang et al., 2016). Spe- bacteria, marine invertebrates and fish (Maki et al., 2001; Halton cifically, aromatic hydrocarbons were photo-oxidized the fastest, et al., 2010) relative to the original oil that had not been aged by followed by n-alkanes and then the weathering-resistant sunlight. biomarker steranes and terpanes. Oil properties, temperature and GC-based techniques are limited to compounds having the solar intensity were other factors affecting the rate of photolysis for volatility below ca. 400 C. Therefore, only some petroleum hy- these petroleum hydrocarbons. However, this previous work did drocarbons (e.g., the GC-detectable petroleum hydrocarbons (TPH), not investigate the fate and behavior of intermediates generated n-alkanes, polycyclic aromatic hydrocarbons (PAHs) and their through photo-oxidation. As aged brine tends to be more toxic due alkylated congeners (APAHs), and some petroleum biomarkers) can to the higher polarity and bioavailability of these oxygenated in- be analyzed using GC-based techniques (D'Auria et al., 2008; termediates compared to PAHs (Maki et al., 2001; Halton et al., D'Auria et al., 2009; John et al., 2016; Yang et al., 2016). Other 2010), it is important to study their fate and behavior related to than the GC-detected targets, most of newly formed oxygenated photo-oxidation to improve our understanding of the environ- intermediates produced through biodegradation/photo-oxidation mental implications of photo-oxidized diluted bitumen. have a low volatility and high polarity due to the addition of oxy- Solar irradiated samples collected at pre-determined time gen. If the profile of these oxygenated components in both raw points were run through the appropriate pre-treatment proced- petroleum products and weathered oils (e.g., from an oil spill) can ures, analyzed by GC-MS for carbonyl PAHs and by high perfor- be measured, it would be possible to gauge the extent of weath- mance liquid chromatography-ultrahigh resolution Orbitrap mass ering that has occurred and, in part, confirm the mechanisms spectrometry combined with electrospray ionization in negative causing this weathering. Studying the variability of these oxygen- mode (HPLC-HRMS) for NAFCs. The chemical composition of ated components within petroleum mixtures will provide a better representative carbonyl PAHs and NAFCs produced through photo- understanding of the environmental weathering processes that oxidation was investigated. This study provides a molecular snap- have occurred over time, and thus, a better understanding of the shot of the compositional changes of oxygenated components in fate and behavior of oil spill events. dilbits at different solar exposure times and for different seasons Oxygenated PAHs are a class of photoproducts. These include (different solar irradiance intensities and environmental tempera- PAH ketones and PAH quinones, formed through the photo- ture). It is important to note that, as of yet, no certified reference transformation of PAHs, which may be subjected to further materials for identifying and quantifying NAFCs exist due to their mineralization or degradation. These products are more toxic than extremely complex chemical composition. Average mass spectra their parent compounds (Mallakin et al., 1999). The other major were generated to identify NAFCs based on the elemental compo- polar photoproducts are naphthenic acid fraction compounds sition of each sample. Therefore, all values reported in this study (NAFCs), which contain a diverse family of carboxylic acids and must be regarded as being semi-quantified as the reported results other acid-extractable organic compounds having varied structures represent the relative abundance normalized to the response of the (e.g., aromatic, adamantane or diamondoid structures) and internal standard for each injection. different species (such as sulfur- and nitrogen-containing compounds and multiple oxygenated acids) (Rowland et al., 2011). 2. Materials and methods NAFCs have been reported as the principal toxicants in oil sands processed waters (OSPW) (Clemente and Fedorak, 2005; Frank Most of the chemicals and standards were supplied by Sigma- et al., 2008, 2009; Vanderveen et al., 2017). Due to their complex Aldrich (Bellefonte, PA, USA) and Sigma-Aldrich (Oakville, ON, composition and limited availability of authentic standards, iden- Canada). A summary is as follows: the LC/MS purity chemicals tifying and quantifying NAFCs is quite limited using conventional include acetic acid, formic acid and ammonium acetate; HPLC grade identification technology. Currently, some advanced analytical solvents include methanol, isopropanol and acetonitrile; deuter- 2 techniques have characterized polar and nonvolatile compounds ated carbonyl PAHs include [ H6] 1,4-naphthequinone (1,4- 2 produced through photochemical transformation (Headley et al., naphthoquinone-d6), and [ H8]9-fluorenone (9-fluorenone-d8); 2 2 2011; Islam et al., 2013; Ray et al., 2014; Vaughan et al., 2016), deuterated fatty acids include [ H15] caprylic acid (C8:0-d15), [ H2] 2 biodegradation (Hughey et al., 2007), natural weathering (Lemkau palmitic acid- (C16:0-d2), and [ H27] myristic acid (C14:0-d27); ten et al., 2014), or found in some crude oils (Stanford et al., 2007) and carbonyl PAH standards include 1-indanone, 1,4-naphthoquinone, aqueous matrices (Brunswick et al., 2015). Ultrahigh resolution acenaphthenequinone, 9-fluorenone, peinaphthenone, anthraqui- electrospray ionization in negative mode (ESI-) Orbitrap mass none, 9, 10-phenanthrenequinone, 2-ethyl anthraquinone, benzo spectrometry (ESI() Orbitrap MS) has been shown to be suitable [a]fluorenone, and 5, 12-nanphthacenequinone All other solvents for the characterization of naphthenic acids with monocarboxylic used are of the highest purity and obtained from Caledon (Geor- acid (O2-NAs) and NAFCs in Athabasca Basin OSPW (Headley et al., getown, ON, Canada). Ultra-pure water was prepared in-house 2012). Therefore, this technique could be used to identify and from a Milli-Q water purification system (Millipore, Billerica, MA, quantify NAFCs within solar irradiated petroleum oils, and to USA). Silica gel (100e200 mesh) was supplied by Spectrum evaluate the fate and behavior of NAFCs produced through photo- Chemicals (Gardena, CA, USA). oxidation at a molecular level. This study is an extension of our earlier work (Yang et al., 2016) 2.1. Exposure experiments where two diluted bitumen oils, Cold Lake Blend (CLB) and Accessed Western Blend (AWB), and one conventional light crude Stock solutions of the CLB and AWB dilbits and one represen- oil (Alberta Sweet Mixed Blend, ASMB), were spiked onto the sur- tative light crude oil (ASMB) were prepared by dissolving in hexane face of saltwater and irradiated with natural solar light (Ottawa, and several drops of dichloromethane (DCM) to 150 mg/mL. The

100 624 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634 dilbit products used in this study were relatively viscous. A mass of 2.2. Analytical procedures 300 mg of oil was used for the photolysis study. Solvent dilution is necessary to quantitatively spike the oil onto the water phase. All solar irradiated and control samples were spiked with the Hexane was selected based on its stability as an oil solvent and appropriate surrogates (9-fluorenone-d8, C8:0-d15 and C14:0-d27) density <1 g/mL. Hexane would float on the water surface, rapidly and acidified with hydrochloric acid to a pH value of 2 to maxi- evaporate to air, leaving the oil on the water surface and available mize the extraction efficiency. They were extracted in triplicate for photo-oxidation by solar irradiation. DCM is necessary to ensure using dichloromethane (DCM). The combined extracts were the complete dissolution and homogeneity of CLB and AWB in their concentrated to 10 mL. Appropriate amounts of extracts were sol- final stock solutions. vent exchanged to hexane prior to column fractionation for The detailed physicochemical information for the three test carbonyl PAH analysis. The solvent exchanged extracts were loaded oils and the solar irradiation conditions were presented in our into a column (10 mm i.d.) with 3 g of silica gel topped with 1 cm of previous report (Yang et al., 2016). In brief, the two dilbits, espe- anhydrous Na2SO4. Following the elution of aliphatics with hexane cially for AWB, contain higher resins and asphaltenes but have less and aromatics with a mixture of hexane and DCM (1:1, v: v), a 15- resolved peaks than ASMB (accounting for 35% of TPH in ASMB). mL mixture of DCM and acetone (95:5 of v: v) was used to elute the This suggests that the two dilbits contain more of the heavier carbonyl PAHs. The eluents were then dried by gentle flow of N2 compounds than ASMB. and spiked with d14-terphenyl as the internal standard for carbonyl Saltwater was prepared by dissolving sodium chloride into PAHs analysis. For the analysis of NAFCs, the appropriate extracts ultra-pure water to a final concentration of 0.33% (mass/mass) and were loaded into a column (5 mm i.d.) with 1 g of silica gel topped used to simulate a seawater environment. Then 200 mL of salt- with 1 cm of anhydrous Na2SO4. Ten millilitres of DCM were used to water was poured into a 400-mL glass beaker and 2 mL of the remove most of the non- to medium-polar oil components to above oil stock solution was spiked onto the surface of the arti- minimize background interferences for the target analysis. Another ficial brine. A time of 30 min was allotted for the hexane solvent to 7 mL of methanol, containing 0.1% formic acid, was used to elute the evaporate from the oil spiked onto the water surface. Preliminary polar targets. The eluents were then dried by gentle flow of N2 and work showed that 30 min was twice the time required for a similar reconstituted in 1 mL of HPLC grade isopropanol with 1.0 mg/mL of volume of hexane alone to evaporate from the surface of water. deuterated palmatic acid (C16:0-d2) as the internal standard. All Three replicates were prepared for each sample, two were samples were centrifuged at 3000 rpm for 10 min prior to HPLC- exposed to sunlight, and one acted as a control. Both exposed HRMS analysis. samples were covered with a polyethylene membrane to protect the samples from rain, snow and other materials, and to minimize 2.3. Instrumental analysis evaporation loss of the volatile oil components. It is noted that the polyethylene membrane can screen off a third of the illuminance 2.3.1. GC-MS analysis of carbonyl PAHs of UV-B. To account for this, the sensor used to measure UV-B Characterization of carbonyl PAHs was performed on an Agilent intensity was also covered with a similar polyethylene mem- 6890 GC system interfaced to an Agilent 5973 mass spectrometer. brane. A control sample was completely wrapped with aluminum The target compounds were separated on a DB-5 MS capillary foil. It was kept in the same environment as the test vessels and column (30 m 0.25 mm i.d., 0.25 mm film thickness) with the ruled out the potential influence of other environmental variables following temperature program: 60 C for 2 min, heated to 300 C such as temperature and humidity. The three replicate samples at 6 C/min and held for 20 min at 300 C. For all GC/MS analyses, were then placed outside on a rooftop of a two-story building ultrahigh purity helium was employed as the carrier gas at a flow located in Ottawa, Ontario. Two series of studies were conducted rate of 1 mL/min. The temperatures of injector, transfer line, ion during the winter (December 18, 2014 to March 19, 2015) and source and MS quadrupole analyzer were held at 290, 290, 300, and summer (May 13 to August 11, 2015) periods respectively. One 150 C, respectively. Samples were injected in a splitless/split mode. control and two irradiated samples were removed from the The mass-selective detector (MSD) was operated at an electron rooftop after 5, 15, 30, 60 and 90 days, respectively, for subsequent impact mode (70 eV) for selected ion monitoring (SIM) runs. laboratory analysis. Temperature was continually monitored throughout the 90- 2.3.2. HPLC-HRMS analysis of NAFCs day exposure. Each day at around 1 p.m., a UV-B meter was used The analytical methods were adapted from Zhang et al. (2014). to measure UV-B intensity as UV-B provides the primary energy in In brief, sample analysis was performed on an Accela HPLC system the solar spectra for photo degradation and UV-B intensity paired with Exactive Orbitrap Mass System (Thermo Fisher Scien- changes depending on the weather. Weather during the winter tific, San Jose, CA, USA) and electrospray ionization (ESI) source in exposure study was typical for the location and varied from clear negative ion mode. A 20-min LC gradient method using a Poroshell and sunny days to periods of strong winds and heavy snowfall. 120 EC-C8 column (100 2.1 mm i.d., 2.7 mm film thickness from The winter temperature and the UV-B intensity ranged from 30 Agilent) at a flow rate of 0.25 mL/min was used for separation. The to 8 C, and 0.5 to 3 mw/cm2, respectively. Any accumulation of gradient method was programmed among three mobile phases: snowfall was removed from the test vessels each morning and mobile phase A, water with 0.0025% formic acid; mobile phase B, every two hours during the day to maximize the exposure of the 95% acetonitrile/5% water with 3 mmol/L ammonium acetate; and samples to daylight. Ice was occasionally formed in the test ves- mobile phase C, 85% isopropanol/10% toluene/5% water with sels during the winter exposure. Summer weather was also typical 3 mmol/L ammonium acetate. In the beginning, the 95% mobile of the location included extensive periods of clear and sunny phase A and 5% mobile phase B were held isocratically for 2 min, conditions as well as days of significant wind and rain. Summer followed by a linear increase of mobile phase B from 5 to 100% over temperatures varied between 3and34C, while UV-B intensity 8 min. Mobile phase B decreased to 10% and C increased to 90% in a ranged between 3 and 15 mw/cm2. The average hours of sunlight linear fashion over the following 5 min and then held for 10 min. throughout the 90-day exposure were estimated at 300 Then mobile phase B increased to 100% over 0.5 min and held this (3.3 h$day 1) in winter and 745 h (8.7 h$day 1)insummer, level for 1.5 min. Mobile phase B then decreased to 5%, and mobile respectively. phase A increased to 95% over 0.5 min, and held these levels for 5 min prior to the next injection. The Orbitrap mass spectrometer

101 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634 625 was operated in ESI negative mode (ESI-) and data was acquired in samples. For example, anthraquinone, 9, 10-phenanthrenequinone, full scan from the mass to charge (m/z) ratio of 80e1600. The 2-ethyl anthraquinone, benzo[a]fluorenone, 5, 12- negative ion of the acetic acid dimer (2 M-1, m/z ¼ 119.03498) was naphthacenequinone, fluorenone and a mixture of fluorenones as used as lock mass for scan-to-scan calibration correction, to ensure well as phenanthrenequinones having different levels of alkylation less than 2 ppm mass accuracy. The mass parameters were set as: were identified in most of the irradiated samples. Fig. S1 depicts the spray voltage, 4.0 kV; capillary temperature, 320 C; sheath gas, 45; chromatograms of fluorenone and the mixture of fluorenones and auxiliary gas, 15; and tube lens, 90 V. phenanthrenequinones with different levels of alkylation for a Heteroatom containing species including the formula of representative sample. Limited available standards were used for CcHhNnOoSs, where c, h, n, o and s are the elemental numbers in identifying some of the carbonyl PAHs in this study; a literature formula (0 c 100, 0 h 202, 0 o 8, and 0 n, s 2), were study was used to identify the carbonyl PAHs without authentic detected as deprotonated molecules by the Orbitrap MS ESI-scan, standards available. The m/z ¼ 180 was identified as fluorenone, which are [M-H]- ions. Spectral interpretation of NAFCs was ob- the photoproduct of fluorene; m/z ¼ 194 may be the result of tained to retrieve the average mass spectrum list across a retention photo-oxidation of C1-fluorenes to form C1-fluroenones; m/z ¼ 208 time span of 5e20 min. Elemental formulae were generated at a may be a mixture of anthraquinone, 9,10-phenanthrenequinone mass accuracy of 2 ppm and, based on m/z values, assigned by and/or C2-fluorenones from the photo-oxidation of phenan- Xcalibur software package (Thermo Fisher Scientific, San Jose, CA, threne/anthracene and C2-fluorenes; m/z ¼ 222 may be a mixture USA). The MS data was processed and the elemental composition of of C3-fluorenones and/or C1-phenanthrenequinones from the the compounds was determined by measuring the accurate m/z photo-oxidation of C3-fluorenes and C1-phenanthrenes; m/z ¼ 236 values. A resolving power (m/Dm50%) of ~100,000 and a mass ac- may be a mixture of C4-fluorenones and/or C2- phenan- curacy of <2 ppm provided unambiguous molecular formula as- threnequinones from the photolysis of C4-fluorenes and/or C2- signments for singly charged molecular ions. Double bond phenanthrenes. equivalent (DBE) values represent the number of rings plus the Plots of the quantified representative carbonyl PAHs formed number of carbon double bonds in a given molecular formula, through photolysis in relation to solar exposure time for summer which is calculated by Equation (1) for the elemental formulae of and winter seasons are shown in Fig. 1. With summer exposure, all CcHhNnOoSs (McLafferty and Turecek, 1993): the carbonyl PAHs were formed at the beginning of solar exposure but were subsequently removed by continued exposure. These h n fi DBE ¼ c þ þ 1 (1) ndings indicate that these oxygenated PAHs formed through the 2 2 photo-oxidation of their parent PAHs, were further photo-oxidized Isotopes, adducts and minors were excluded from NAFC anal- or mineralized. Once their rates of formation were less than their ysis. Because there are no available weathering-resistant bio- rates of photolytic degradation, a decrease in their abundance was markers for the Orbitrap MS to permit normalization across the observed. dataset, the responses for m/z peaks based on the assigned mo- The test oils exhibited different patterns of abundance because lecular formulae were normalized to the abundance of the internal the formation rate of the compounds was dependent on the photo- standard for each mass spectrum. This normalization can partially transformation rates of their precursors. The highest amount of eliminate detection bias in the ESI response caused by sample carbonyl PAHs was observed after a 5-day exposure of ASMB in the matrix effects and/or ionization suppression effects, as ionization summer and after 15 days for CLB and AWB in the summer season. fi suppression will cause an apparent reduction in analyte recovery A signi cant decreasing trend of the formed carbonyl PAHs was also from different matrices when ESI in negative ion mode is used noted for ASMB (e.g., after a 5-day exposure in summer) and CLB (Page et al., 2007). Assuming the ionization efficiencies of the tar- then AWB (e.g., after 15-day exposure in summer), suggesting that gets across samples are similar after the normalization above, the their photolysis rates were faster than their formation rates after fl observed alteration over exposure time, season of exposure and oil the initial period of irradiation. Among the three test oils, uo- type will highlight any trends related to solar exposure. renones and/or phenanthrenequinones were most abundant in ASMB, followed by CLB and then AWB, after summer irradiation. This is consistent with the photolytic rates of fluorenes and phen- 2.4. Quality control and quality assurance anthrenes that reported in one of our previous studies (Yang et al., 2016). The photolytic rates of these precursors were highest in the Ultra-pure water spiked with same amount of salt and surro- summer for ASMB, followed by CLB and then AWB. This is because gates as presented above was prepared for blank control test. The the physicochemical properties of ASMB are significantly different same analytical procedures were run for the blank controls to from the two diluted bitumen products. For example, ASMB con- evaluate any background contribution to NAFCs analysis. All the tains more resolved peaks than CLB and AWB. Resolved peaks are reported NAFC data were background corrected in this study. The usually easier to photo-oxidize or biodegrade compared to unre- ± ± surrogate recoveries were determined to be 85 15%, 67 13% and solved complex mixtures (Wang et al., 2016). The low viscosity of ± fl 98 20% for uorenone-d8, C8:0-d15 and C14:0-d27, respectively, ASMB also increased exposure to solar light due to a larger when compared to the response of the authentic standards. spreading area than the two dilbits on the surface of water. In the summer season, the fluorenes and/or phenanthrenes in ASMB were 3. Results and discussion extensively photolyzed after a 5-day irradiation and the formed photoproducts were subsequently photo-oxidized to other in- 3.1. Variation of carbonyl PAHs produced through photo-oxidation termediates or mineralized compounds by continued solar exposure. The primary photo-oxidation pathway of PAHs proceeds via In the winter season, these carbonyl PAHs in ASMB were formed unstable endoperoxide and/or peroxide intermediates leading to over the first 30 days, then photolyzed by continued exposure. diols and quinones (Nikolaou et al., 1984; Mcconkey et al., 1997). In However, in the CLB, the carbonyl PAHs were continually formed this study, newly formed carbonyl PAHs were identified from the during the 90 days of solar exposure. ASMB also produced more of oils after solar exposure in both summer and winter seasons, these carbonyl PAHs than CLB. It is therefore concluded that oil however these compounds were not detected in any of the control properties are a critical factor affecting the variability of these

102 626 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634

two control samples of the dilbits suggesting that more acidic components were present in the two dilbits than in the ASMB. This finding is reasonable as ASMB contains fewer resins and asphaltenes than CLB and AWB, while bitumen has undergone long-term weathering in the natural environment (Yang et al., 2016). After 30 days of summer irradiation, the numbers of identified peaks increased significantly for ASMB, but not for the two dilbits. A relative shift of the most abundant peaks to a lower mass and an increase in absolute intensity were observed for all the irradiated samples, even though the increased absolute intensity varied for specific oil. For example, the most abundant peak in the irradiated ASMB was about five times greater than its control, whereas CLB was only two times greater than its control. Only a slight increase was observed for AWB. These results indicate that the compositional complexity and/or the contents of NAFCs increased after irradiation, especially for the lower molecular weight congeners in ASMB. An increase in the number of peaks and a higher abundance of acidic components in ASMB both suggest that its photolysis was more thorough than for the two dilbits. This is in agreement with our previous report (Yang et al., 2016) where the target petroleum hydrocarbons in ASMB, especially aromatics, were degraded faster than those in the two dilbits. The irradiated samples appear to have formed abundant acidic compounds having a short carbon chain length (Ray et al., 2014) through the photo-oxidation of non- to medium-polar petroleum hydrocarbons or the intermediates of aldehydes or ketones. Lower molecular weight NAFCs have been reported to be more susceptible to microbial activity and bioavailability (Frank et al., 2008, 2009). Thus, the aquatic toxicity of solar aged oil may be stronger than the original oil itself.

3.3. Van Krevelen diagrams for comparing compound class after solar irradiation

As data were generated by combining the average mass spectra with the assigned molecular formulae of NAFCs in this study, van Krevelen diagrams were used to visualize complex organic samples by comparing the molar ratio of hydrogen to carbon (H/C) and oxygen to carbon (O/C) to major chemical compound classes. These diagrams qualitatively identify the difference among different samples through photo-oxidation. Fig. 2 and Fig. S3 compare the H/ C and O/C ratios for all components that were >0.5% relative to the most abundant peak by HPLC-ESI ()-Orbitrap MS before and after solar irradiation for both seasons. Usually, H/C ratios can be used as aromatic indices to visualize differences between the degree of unsaturation among samples (Koch and Dittmar, 2005). In these diagrams, the H/C differences (due to differences in DBE) are separated vertically. As H/C ratios decrease, the number of rings

Fig. 1. Abundance of representative carbonyl PAHs produced through photo-oxidation plus double bonds increases. In this study, most H/C ratios were for the test oils CLB, ASMB and AWB in two seasons: (a) summer and (b) winter. >1.0 for the three control oils. Multiple congeners with H/C ratios ranging from 0.5 to 2.5 were formed during irradiation. However, some of the NAFCs having lower H/C ratios were decomposed oxygenated PAHs derived through photo-oxidation due to the further after 60e90 days of irradiation. This was particularly the varied photolytic rates of their parent PAHs among different oils. case for the exposed ASMB in the summer. It primarily confirms The duration of solar irradiation, solar intensity and temperature that the formation of NAFCs depends on the specific oil being also control the photolysis rates of PAHs and their alkylated con- irradiated, the season and duration of exposure and the chemical geners (Saeed et al., 2011; Gong et al., 2015; Yang et al., 2016). These structure of NAFCs themselves. factors ultimately affect the fate of these oxygenated PAHs pro- The heteroatom Oo classes (e.g., o from 1 to 8) are clearly duced through photo-oxidation in both seasons. separated horizontally by O/C ratios, which distinguish classes that differ in the number of O atoms. In the present study, most of the O/ 3.2. NAFCs derived from photo-oxidation C ratios were less than 0.4 in the three control samples. Many components having >0.4 O/C ratios were formed after irradiation. It The average negative ion ESI Orbitrap mass spectra of the acidic is obvious that solar irradiation resulted in the oxidation of non- to extractable fractions in the control and 30-day summer irradiated medium-polar petroleum hydrocarbons or some intermediates by test oils are depicted in Fig. S2. The number of the identified peaks adding oxygen into them, producing abundant oxygenated com- in the control sample of ASMB was far less than observed for the ponents with higher water solubility and bioavailability (Islam

103 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634 627

Fig. 2. Van Krevelen plots for the total ions of the acid fraction analyzed by () ESI-Orbitrap-MS for the representative summer control and 30-day, 60-day and 90-day exposed samples for three test oils. et al., 2013; Ray et al., 2014). Given the possible interference from sample based on their heteroatom content by using the assigned the background and the varied ionization efﬁciency for different molecular formulas for each m/z peaks. Compositional trends samples, semi-quantitative results corrected by blank controls and among samples were visualized with heteroatom class distribution normalized by the response of internal standard will be discussed graphs. Fig. 3 shows the heteroatom class distribution for selected in the following sections. species having a >0.1% abundance of the internal standard in each injection in the ESI ()-Orbitrap MS for the control and solar irra- 3.4. Heteroatom class distribution through photo-oxidation diated samples (from control to 90 days) in both seasons. The most abundant acidic heteroatom class for all the control samples were 3.4.1. Total heteroatom class distribution as a function of irradiation components containing the heteroatom of oxygen only (Oo species, time components with the formula CcHhOo, where o ranges from 1 to 8). In this study, compounds were grouped into “classes” within a Among the three test oils, AWB had the highest abundance of Oo

104 628 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634

Fig. 3. Relative intensity of the major chemical classes observed of the control and solar exposed samples of test oils for two seasons: (a) summer and (b) winter.

species, followed by CLB and then ASMB. A similar trend was also ﬁnal abundance was far less than that of the Oo species. The for- observed for the other sulfur- and nitrogen-containing species (SOo, mation of NAFCs can be ascribed to the preferential photo- S2Oo, NOo, N2Oo and SsNnOo, where s and n span from 1 to 2). oxidation of aliphatic, cyclic or aromatic hydrocarbons to alde- Signiﬁcant amount of Oo species were formed after solar irra- hydes or ketones, further hydroxylated to carboxylic acid (O2) in- diation in both seasons. Although other species were formed, their termediates and even ring opening for aromatic intermediates (Yue

105 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634 629 et al., 2016). The Oo species were the most abundant components in all the exposed samples. For the three test oils, the N2Oo species were the second most abundant group, followed by the NnSsOo group and then the other species in all the solar exposed ASMB samples. In all the exposed CLB and AWB, SOo was the second most abundant group, followed by the NOo group and then the others. It is evident that the level of original sulfur- and nitrogen-containing components in the three oils determined the level of the NnOo, SsOo and NnSsOo species formed through photo-oxidation. The formation rates of these acidic components were highest in ASMB in both seasons while the two dilbits had a similar formation rates. As with the variation of carbonyl PAHs, the varied formation rates of these acidic components in different oils can be ascribed to the different physicochemical properties of the oils. For example, the photo-oxidation of the non- to medium-polar oil components or some intermediates in ASMB was much faster than for the two dilbits. This is because ASMB is lighter, exposing it to sunlight with a larger surface area; ASMB also contains more resolved peaks, which makes them photo-oxidize by sunlight more easily (Yang et al., 2016). The faster photo-oxidation rates of the precursors resulted in the faster formation rates for these acidic intermediates. When exposed in the summer, the total Oo species were continually formed until 30 days for all three oils. Continual solar exposure did not produce any signiﬁcant changes in the total Oo species of ASMB and CLB until 90 days. Similar results were seen for the S2Oo and N2Oo groups. The continual exposure caused the decomposition of most groups in the AWB and some of the groups in the CLB (SOo and NOo) and ASMB (SOo, NOo and NnSsOo). Most of the groups were continually formed during the 90-day irradiation during the winter exposure. All the species after summer exposure were more abundant than those formed in winter. As discussed previously, summer exposure had a longer sunlight irradiation duration, higher solar intensity and warmer temperature than the winter exposure. All these factors contributed to the faster photo- oxidation of the non- to medium-polar oil components, thus producing more acidic photoproducts in summer than in winter (Yang et al., 2016). Thus, the intensity and duration of solar irradiation and air/water temperature are some of the main factors affecting the fate of these heteroatom species produced through photo- oxidation.

3.4.2. Oo acidic class as a function of irradiation time Monocarboxylic acids (O2 species) are the most abundant oxygen-containing acidic components in crude oil and diluted bitumen (Colati et al., 2013). However, solar irradiation can cause the addition of oxygen into oil components to form higher-order Fig. 4. Relative intensity of Oo series observed of the control and solar exposed sam- oxygen classes (O4O7)(Ray et al., 2014; Vaughan et al., 2016). As ples of test oils for two seasons: (a) summer and (b) winter. such, the Oo species discussed earlier may comprise different oxygen numbers. Herein, Oo species having different oxygen numbers duration of irradiation was also dependent on oxygen number. All (from 1 to 8) were subdivided from the total Oo species to evaluate sub-O species increased during the winter irradiation period, even their variation through photo-oxidation. Fig. 4 compares the rela- o though their rates of increase were far less than those observed in tive abundance for O1O8 acidic species in the control and irradi- the summer. All these findings further demonstrated that photo- ated samples. In samples without irradiation (control oils), O2 was chemical reactions increased the number of oxygen atoms per the most abundant class. Following irradiation, the O2 class and the molecule thus increasing the water solubility of oil-derived com- other higher-order oxygen classes (O3O8) increased significantly pounds. This is in agreement with a previous study looking at the in abundance, while the abundance of O1 class did not undergo any Deepwater Horizon oil (Ray et al., 2014). Therefore, highly marked changes. The relative abundance of all the O2 to O8 species bioavailable and toxic photoproducts were formed through the increased until the 30-day point for irradiation in the summer. photo-oxidation of oil components (King et al., 2014); however, the Then, O2 species decreased, but all the other higher-order Oo oxidation mechanism is difficult to address due to the complexity of (O3O8) species did not change significantly with the continual oil components. irradiation. Similar to the observations mentioned in the previous sections, summer exposure produced more abundant O2 to O8 species than winter exposure and ASMB had a higher abundance of 3.4.3. SsOo and NnOo acidic classes as a function of irradiation time O2 to O8 species than did CLB and AWB in both seasons. Thus, the Figs. S3 and S4 depict the variation trends of individual SsOo and variable abundance of the different sub-Oo species related to the NnOo acidic species in all the test oils. It is noted that the SsO7, SsO8,

106 630 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634

NO7, NO8, N2 and N2O groups were not shown in these two figures irradiation (30 days in the winter and 15 days in the summer) while as their abundance are relatively low compared to the other SsOo some were reduced in abundance with further irradiation. The two- and NnOo species. As mentioned previously, both acidic species dimensional plots of the subtotal NAFCs versus DBEs/carbon were far less abundant than the Oo species in both the control and numbers provide further information regarding their behavior due irradiated samples. In the control samples, some of the SOo species, to photo-oxidation along the degree of unsaturation or carbon e.g., SO2, SO3 and SO4, were more abundant than S2Oo species. The chain length. two dilbits had a higher level of SsOo species than ASMB. The solar irradiation caused the formation of most of SsOo species in the first 3.5.1. Relative intensity of Oo species versus DBE produced through 5e15 days of exposure, including SO2, SO3, SO4,S2O, S2O2, S2O3 and photo-oxidation S2O4. However, continual exposure then produced a decrease in The plots of the relative abundance of subtotal Oo species versus SsOo species in both seasons. It seems that season did not signifi- DBEs (from 1 to 16, corresponding naphthenic rings from 0 to 15) cantly affect the production of these SsOo species. The two dilbits are shown in Fig. 6. The abundance of all the Oo species decreased produced more SsOo species than ASMB. This is possible because with the increase of DBE values or cyclic rings in the control ASMB. dilbits had abundant sulfur-containing PAHs (e.g., dibenzothio- The Oo species with a DBE of 3 (2 cyclic rings) were the most phenes and benzonaphthothiophenes) (Yang et al., 2016) and the abundant group in the control AWB and CLB, followed by those photochemical reactions of these parent sulfur-containing PAHs with DBE ¼ 2 (one cyclic ring) and 1 (acyclic ring). An overall produced these SsOo species (Bobinger and Andersson, 2009). decreasing trend was then observed as DBE increased. After irra- The ASMB control sample had a greater abundance of NOo than diation, the most abundant group had a DBE of 2 (i.e., 1 double bond N2Oo species than the two control dilbits. NO, NO2 and NO3 were the or one cyclic ring), followed by the acyclic group, and then the most abundant classes in ASMB control, while the N2O4 species DBE ¼ 3 and 4 congeners. All the other groups with >5 DBEs were the most abundant class for the two control dilbits, followed generally decreased as DBE values increased. However, species by N, NO, NO2 and NO3 species. After irradiation, more heavier- having a DBE ¼ 6 (5 cyclic rings or double bonds) and DBE ¼ 11 (10 order N2Oo species (e.g., N2O5N2O6) and NOo species (from NO2 cyclic rings or double bonds) were more abundant than their to NO6) were produced by all three oils in both seasons. The relative nearby DBE species in most of the oils. This phenomenon suggests distribution (normalized to 100% for all NnOo species) of NnOo that these Oo species may be derived from the photolysis of some species shifted from left to right, in agreement with an increased hydrocarbons having polycyclic aromatic rings because these pre- oxygenation after irradiation. However, a detailed analysis of the cursors are the most photo-sensitive group (Yang et al., 2016). variability of these individual species was not undertaken as the Congeners having lower DBE values (e.g., from 1 to 4) increased abundance of SsOo and NnOo species is relatively low compared with at the beginning of irradiation in the summer, but then either Oo species. Their variation due to the duration of the irradiation decreased or were unaltered with further irradiation in the case of may cause bias due to the potential measurement uncertainties. the two dilbits or they continually increased as in the ASMB. A significant decreasing trend was observed for those species having 3.5. Compositional variation of major acidic species produced a higher number of naphthenic rings (e.g., DBE ¼ 5e10) ASMB and through photo-oxidation AWB (but not CLB) after 30 days of irradiation in the summer. This can be partially ascribed to the following possibilities. First, it seems All the data discussed above are based on the total or subtotal of that the photo-stability of Oo species depends on their DBE values. the three major acidic species. As mentioned in Section 2, the Generally, as the DBE value increases, the photo-stability of Oo elemental composition of these acidic species varies with different species decreases. NAFCs having a greater number of naphthenic carbon, sulfur, oxygen and nitrogen numbers as well as the rings may be more photo-sensitive than those having fewer rings different number of double bond/naphthenic rings. Each ring or (King et al., 2014; Leshuk et al., 2016), which would result in a double bond in a molecule reduces the number of hydrogen atoms reduced detection level. It was reported that the removal of NAFCs by two (Lemkau et al., 2014). DBE, representing the number of rings having one naphthenic ring was slower than for NAFCs having a plus double bonds, is a very useful parameter for examining the higher number of rings (>1) when an advanced oxidation of compositional differences across samples. As Oo are the most naphthenic acids via ozone treatment was used to treat OSPW abundant species among the three major acidic groups, the varia- (Gamal et al., 2011). This pattern is opposite to the biodegradation tion of the elemental composition of Oo species from O1 to O8 of NAFCs, where those characterized by more cyclic rings tend to be through photo-oxidation was compared by colour-coded iso- the most environmentally persistent and most resistant to abundance contour plots of DBE versus carbon number (Fig. 5). biodegradation (Misiti et al., 2014). Second, in the summer period, These three-dimensional plots highlight compositional changes ASMB was photolyzed the fastest among the three oils (Yang et al., that occurred over the 90 days of solar exposure to provide a 2016) and the depletive consumption of the photo-sensitive pre- comparison of the compositional changes within a heteroatom cursors could not make up the loss due to the photolysis of these class. formed acidic components. Finally, the photolysis rates of the non- Analyzing the colour codes, all solar irradiated samples had a to medium-polar oil components or intermediates in AWB were the higher abundance of Oo species than their corresponding controls. lowest among the three oils. As such, their formation rates were The colour spectrum became wider in both the vertical and horizon limited relative to their photolysis rates. It can be concluded that axes following irradiation, suggesting the production of more acidic abundant NAFCs have been formed through photo-oxygenation of Oo species. The most abundant Oo species in all the irradiated oil components, while those having a high degree of saturation are samples shifted to the left in the horizontal direction compared the most abundant groups in oils aged by sunlight. with all the control samples. This suggested that the newly formed Oo species were abundant for congeners having lower carbon 3.5.2. Relative intensity of Oo species versus carbon numbers numbers. through photo-oxidation Among the different oils, the shape of the colour plots was The plots of the relative abundance of subtotal Oo species versus flatter vertically in CLB and AWB than ASMB for all the exposed carbon number (from C6 to C60) are shown in Fig. 7. A typical bell samples. It seems that the solar irradiated oils (especially for ASMB) shape with the maximal species in the carbon range of 16e24 was formed abundant NAFCs with >4 DBEs at the beginning of solar shown for all the controls. Abundant Oo species were produced in

107 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634 631

Fig. 5. Isoabundance contour plots of the double bond equivalent (DBE) vs. carbon number for acidic species derived from negative ion ESI Orbitrap mass spectra for Oo classes in the control and solar exposed samples (after a 5-, 15-, 30-, 60- and 90-day exposure period) in summer (ﬁrst three columns on the left-hand side) and winter (two columns on the right-hand side).

the carbon range of less than C50, C30C40, and C25 in ASMB, CLB and oxygenation of oil compounds or the further oxygenation of the AWB, respectively. No significant change was observed for those formed NAFCs increased the water solubility of the photoproducts, species being > C50 in all the oils. It can be concluded that solar thereby increasing their potential bioavailability and aqueous irradiation produced abundant Oo species within specific carbon toxicity (King et al., 2014; Ray et al., 2014). range for different oils. The most abundant Oo species shifted from right to left on the plots following solar irradiation (e.g., from C24 to 4. Conclusions C16 for AWB, C21 to C16 for CLB, and C20 to C15C16 for ASMB). Therefore, solar irradiation produced abundant NAFCs having a The fate of identified oxygenated PAHs and polar acidic fi shorter carbon chain length. These ndings are in agreement with extractable fractions with the elemental composition of CcHhOoSsNn previous reports for water-soluble acidic extracts detected after the was investigated by exposing diluted bitumen to natural sunlight in solar irradiation of crude oil (Ray et al., 2014). As mentioned above, saltwater. Various carbonyl PAHs were identified. Fluorenone and the formation of the petroleum precursors and photolysis of NAFCs the mixture of fluorenones and/or phenanthrenequinones were the occurred simultaneously during the irradiation process. The left most abundant congeners identified. Abundant acidic components shift of the maximal NAFCs can be partially attributed to the pref- having the most abundant species of Oo (from O2 to O8)were erential degradation of oil components or NAFCs having higher formed after irradiation, as were SsOo and NnOo species, even carbon numbers as they are easier to interrupt during photo- though their level was far below that of the Oo class. The dominant oxidation, especially for junctions that are far away from the ring Oo species after solar irradiation shifted to a lower carbon number, a (Maki et al., 2001; Liu et al., 2016). It is clear that either the direct higher degree of saturation and higher oxygen numbers. These

108 632 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634

Fig. 6. Relative intensity of subtotal Oo species versus DBE observed for the control and solar irradiated samples for two seasons: (a) summer and (b) winter. intermediates were generally produced faster in light crude oil and ecological effects of oil aged in sunlight following a spill depend on summer exposure compared with the two dilbits and winter the speciﬁc oil and the environmental conditions. Given the exposure. The fate of these oxygenated components derived marked increase in water solubility of these oxidized in- through solar irradiation depends on oil properties and chemical termediates, they may pose an enhanced toxicity to aquatic or- structure of the intermediates themselves, the duration of expo- ganisms during the photolysis process. sure, solar intensity and temperature. This implies that the

109 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634 633

Fig. 7. Relative intensity of subtotal Oo species versus carbon number observed for the control and solar irradiated samples for two seasons: (a) summer and (b) winter.

110 634 Z. Yang et al. / Environmental Pollution 231 (2017) 622e634

Acknowledgements when re-exposed to sunlight. Sci. Total Environ. 573, 189e202. King, S., Leaf, P., Olson, P., Ray, M., Tarr, M., 2014. Photolytic and photocatalytic degrada-tion of surface oil from the Deepwater Horizon spill. Chemosphere 95, This work was funded and supported by the Government of 415e422. Canada's World Class Tanker Safety System (WCTSS) program and Koch, B.P., Dittmar, T., 2005. From mass to structure: an aromaticity index for high- Ocean Protection Plan program. Oni Olatunji and Suhash Aravindan resolution mass data of natural organic matter. Rapid Commun. Mass Spectrom. 20, 926e932. are greatly thanked for editing the programs used for data analysis. Lemkau, K.L., McKenna, A., Podgorski, D.C., Rodgers, R.P., Reddy, C.M., 2014. Mo- lecular evidence of heavy-oil weathering following the M/V cosco busan spill: Appendix A. Supplementary data insights from fourier transform ion cyclotron resonance mass spectrometry. Environ. Sci. Technol. 48, 3760e3767. Leshuk, T., Wong, T., Linley, S., Peru, K.M., Headley, J.V., Gu, F., 2016. Solar photo- Supplementary data related to this article can be found at http:// catalytic degradation of naphthenic acids in oil sands process-affected water. dx.doi.org/10.1016/j.envpol.2017.08.043. Chemosphere 144, 1854e1861. Liu, J., Wang, L., Tang, J., Ma, J., 2016. Photocatalytic degradation of commercially sourced naphthenic acids by TiO2-graphene composite nanomaterial. Chemo- References sphere 149, 328e335. Maki, H., Sasaki, T., Harayama, S., 2001. Photo-oxidation of biodegraded crude oil Aeppli, C., Carmichael, C.A., Nelson, R.K., Lemkau, K.L., Graham, W.M., and toxicity of the photo-oxidized products. Chemosphere 44, 1145e1151. Redmond, M.C., Valentine, D.L., Reddy, C.M., 2012. Oil weathering after the Mallakin, A., McConkey, B.J., Miao, G., McKibben, B., Snieckus, V., Dixon, D.G., Deepwater Horizon disaster led to the formation of oxygenated residues. En- Greenberg, B.M., 1999. Impact of structural photomodification on the toxicity of viron. Sci. Technol. 46, 8799e8807. environmental contaminants: anthracene photooxidation products. Ecotox. Bobinger, S., Andersson, J.T., 2009. Photooxidation products of polycyclic aromatic Environ. Saf. 43, 204e212. compounds containing sulfur. Environ. Sci. Technol. 43, 8119e8125. Mcconkey, B.J., Duxbury, C.L., Geroge Dixon, D., Greenberg, B.M., 1997. Toxicity of a Brunswick, P., Shang, D., van Aggelen, G., Hindle, R., Hewitt, L.M., Frank, R.A., PAH photooxidation product to the bacteria PHOTOBACTERIUM PHOSPHOR- Haberl, M., Kim, M., 2015. Trace analysis of total naphthenic acids in aqueous EUM and the duckweed LEMNA GIBBA: effect of phenanthrene and its primary environmental matrices by liquid chromatograph/mass spectrometry- photoproduct, phenanthrenequinone. Environ. Toxicol. Chem. 16, 892e899. quadrupole time of flight mass spectrometry direct injection. J. Chromatogr. A McLafferty, F.W., Turecek, F., 1993. Interpretation of Mass Spectra, fourth ed. Uni- 1405, 49e71. versity Science Books, Mill Vallet, CA. Clemente, J.S., Fedorak, P.M., 2005. A review of the occurrence, analyses, toxicity Misiti, T.M., Tezel, U., Pavlostathis, S.G., 2014. Effect of alkyl side chain location and and biodegradation of naphthenic acids. Chemosphere 60, 585e600. cyclicity on the aerobic biotransformation of naphthenic acids. Environ. Sci. ~ Colati, K.A.P., Dalmaschio, G.P., de Castro, E.V.R., Gomes, A.O., Vaz, B.G., RomA£o, W., Technol. 45, 7431e7437. 2013. Monitoring the liquid/liquid extraction of naphthenic acids in Brazilian Nikolaou, K., Masclet, P., Mouvier, G., 1984. Sources and chemical reactivity of crude oil using electrospray ionization FT-ICR mass spectrometry (ESI FT-ICR polynuclear aromatic hydrocarbons in the atmosphere-A critical review. Sci. MS). Fuel 108, 647e655. Total Environ. 32, 103e132. Crosby, S., Fay, R., Groark, C., Kani, A., Smith, J.R., Sulivan, T., Pavia, R., 2013. Page, J.S., Kelly, R.T., Tang, K., Smith, R.D., 2007. Ionization and transmission effi- Transporting Alberta oil sands products: defining the issues and assessing the ciency in an electrospray ionization-mass spectrometry interface. J. Am. Soc. risks. In: US Department of Commerce (Ed.), NOAA Technical Memorandum. Mass Spectrom. 18, 1582e1590. Seattle, WA. Radovic, J., Aeppli, C., Nelson, R.K., Jimenez, N., Reddy, C.M., 2014. Assessment of D'Auria, M., Emanuele, L., Racioppi, R., Velluzzi, V., 2009. Photochemical degrada- photochemical processes in marine oil spill fingerprinting. Mar. Pollut. Bullet tion of crude oil: comparison between direct irradiation, photocatalysis, and 79, 268e277. photocatalysis on zeolite. J. Hazard. Mater 164, 32e38. Ray, P.Z., Chen, H., Podgorski, D.C., McKenna, A.M., Tarr, M.A., 2014. Sunlight creates D'Auria, M., Racioppi, R., Velluzzi, V., 2008. Photodegradation of crude oil: liquid oxygenated species in water-soluble fractions of Deepwater horizon oil. injection and headspace solid-phase microextraction for crude oil analysis by J. Hazard. Mater 280, 636e643. Gas Chromatography with Mass Spectrometer Detector. J. Chromatogr. Sci. 46, Rowland, S.J., Scarlett, A.G., Jones, D., West, C.E., Frank, R.A., Hewitt, L.M., 2011. 339e344. Sterodial aromatic 'naphthenic acids' in oil sands process-affected water: Frank, R.A., Fischer, K., Kavanagh, R., Solomon, K.R., 2009. Effect of carboxylic acid structural comparisons with environmental estrogens. Environ. Sci. Technol. 45, content on the acute toxicity of oil sands naphthenic acids. Environ. Sci. Tech- 9806e9815. nol. 43, 266e271. Saeed, T., Ali, L.N., Al-Bloushi, A., Al-Hashash, H., Al-Bahloul, M., Al-Khabbaz, A., Al- Frank, R.A., Kavanagh, R., Bumison, B.K., Aresnault, G., Headley, J.V., Peru, K.M., Van Khayat, A.E., 2011. Effect of environmental factors on photodegradation of Der Kraak, G., Solomon, K.R., 2008. Toxicity assessment of collected fractions polycyclic aromatic hydrocarbons (PAHs) in the water-soluble fraction of from an extracted naphthenic acid mixture. Chemosphere 72, 1309e1314. Kuwait crude oil in seawater. Mar. Environ. Res. 72, 143e150. Gamal, E.-D.M., Fu, H., Wang, N., Chelme-Ayala, P., Perez-Estrada, L., Drzewicz, P., Stanford, L.A., Kim, S., Klein, G., Smith, D.F., Rodgers, R.P., Marshall, A.G., 2007. Martin, J.W., Zubot, W., Smith, D.W., 2011. Naphthenic acids speciation and Identification of water-soluble heavy crude oil organic-acids, bases, and neu- removal during petroleum-coke adsorption and ozonation of oil sands process- trals by electrospray ionization and field desorption ionization Fourier trans- affected water. Sci. Total Environ. 409, 5119e5125. form ion cyclotron resonance mass spectrometry. Environ. Sci. Technol. 41, Gong, Y., Fu, J., O'Reilly, S.E., Zhao, D., 2015. Effects of oil dispersants on photo- 2696e2702. degradation of pyrene in marine water. J. Hazard. Mater 287, 142e150. Vanderveen, R., Mclnnis, R., Brunswick, P., Shang, D., Headley, J.V., Peru, K.M., Halton, K., Catherine, A.S., Burrows, D.G., Collier, T.K., Scholz, N.L., Incardona, J.P., Hewitt, L.M., 2017. Toxicity of naphthenic acids to invertebrates: extracts from 2010. Natural sunlight and residual fuel oil are an acutely lethal combination for oil sands process-affected water versus commercial mixtures. Environ. Pollut. fish embryos. Aquat. Toxicol. 99, 56e64. 227, 271e279. Headley, J.V., Barrow, M.P., Peru, K.M., Fahlman, B., Frank, R.A., Bickerton, G., Vaughan, P.P., Wilson, T., Kamerman, R., Hagy, M.E., McKenna, A., Chen, H., McMaster, M.E., Parrott, J.L., Hewitt, L.M., 2011. Preliminary fingerprinting of Jeffrey, W.H., 2016. Photochemical changes in water accommodated fractions of Athabasca oil sands polar organics in environmental samples using electrospray MC252 and surrogate oil created during solar exposure as determined by FT-ICR ionization Fourier transform ion cyclotron resonance mass spectrometry. Rapid MS. Mar. Pollut. Bull. 104, 262e268. Commun. Mass Spectrom. 25, 1899e1909. Wang, Z., Yang, C., Yang, Z., Brown, C.E., Hollebone, B.P., Stout, S.A., 2016. Petroleum Headley, J.V., Peru, K., Fahlman, B., McMartin, D.W., 2012. Selective solvent biomarker fingerprinting for oil spill characterization and source identification. extraction and characterization of the acid extractable fraction of Athabasca oil In: Stout, S.A., Wang, Z. (Eds.), Standard Handbook Oil Spill Environmental sands process waters by Orbitrap mass spectrometry. Int. J. Mass Spectrom. Forensics-fingerprinting and Source Identification. Elsevier Inc., London, UK, 345e347, 104e108. pp. 252e270. Hughey, C.A., Galasso, S.A., Zumberge, J.E., 2007. Detailed compositional comparison Yang, Z., Hollebone, B.P., Brown, C.E., Yang, C., Wang, Z., Zhang, G., Lambert, P., of acidic NSO compounds in biodegraded reservoir and surface crude oils be Landriault, M., Shah, K., 2016. The photolytic behavior of diluted bitumen in negative ion electrospray Fourier transform ion cyclotron resonance mass simulated seawater by exposed to the natural sunlight. Fuel 186, 128e139. spectrometry. Fuel 86, 758e768. Yue, S., Ramsay, B.A., Wang, J., Ramsay, J.A., 2016. Biodegradation and detoxification Incardona, J.P., Swarts, T.L., Edmunds, R.C., Linbo, T.L., Aquilina-Beck, A., Sloan, C.A., of naphthenic acids in oil sands process affected waters. Sci. Total Environ. 572, Gardner, L.D., Block, B.A., Scholz, N.L., 2013. Exxon Valdez to Deepwater Hori- 273e279. zon: comparable toxicity of both crude oils to fish early life stages. Aquat. Zhang, G., Yang, C., Liu, Y., Yang, Z., Wang, Z., Landriault, M., Hollebone, B.P., Toxicol. 142e143, 303e316. Lambert, P., Brown, C.E., 2014. Improved approach to quantitatively and quali- Islam, A., Cho, Y., Yim, U.H., Shim, W.J., Kim, Y.H., Kim, S., 2013. The comparison of tatively characterize carboxylic acid isomers in commercial naphthenic acids naturally weathered oil and artificially photo-degraded oil at the molecular mixture by HPLC-high resolution orbitrap mass spectrometer technique. In: level by a combination of SARA fractionation and FT-ICR MS. J. Hazard. Mater. Proceedings of the 37th AMOP Technical Seminar on Environmental Contami- 263 (Part 2), 404e411. nation and Response. Environment and Climate Change Canada, Ottawa, ON, John, G.F., Han, Y., Clement, T.P., 2016. Weathering patterns of polycyclic aromatic pp. 848e869. hydrocarbons contained in submerged Deepwater Horizon oil spill residues

1 Microbial community composition, functions and activities in the Gulf of Mexico, one

2 year after the Deepwater Horizon accident.

4 Etienne Yergeau1, Christine Maynard1, Sylvie Sanschagrin1, Julie Champagne1, David Juck1, Downloaded from

5 Kenneth Lee2,3 and Charles W. Greer1*

7 1National Research Council Canada, Energy Mining and Environment, Montréal, Quebec, http://aem.asm.org/ 8 Canada

9 2Centre for Offshore Oil, Gas and Energy Research (COOGER), Bedford Institute of

10 Oceanography, Fisheries and Oceans Canada, Dartmouth, Nova Scotia, Canada

11 3Commonwealth Scientific and Industrial Research Organization (CSIRO), Australian Resources

12 Research Centre, Kensington, WA, Australia on December 24, 2018 by guest

14 Keywords: Deepwater Horizon, microbial functions, microbial activities, Gulf of Mexico,

15 metagenomics, metatranscriptomics

17 Running title: Microbiology of the GOM one year after the DWH spill

19 *Corresponding author: [email protected]; Tel: 514-496-6182

112 20 Abstract

21 Several studies have assessed the effects of the released oil on microbes, either during or

22 immediately after the Deepwater Horizon accident. However, little is known about the potential

23 longer-term persistent effects on microbial communities and their functions. In this study, one Downloaded from

24 water column station near the wellhead (3.78 km SW of the wellhead), one water column

25 reference station outside of the affected area (37.77 km SE of the wellhead), and deep-sea

26 sediments near the wellhead (3.66 km SE of the wellhead) were sampled one year after the

27 capping of the well. In order to analyze microbial community composition, function and activity,

28 we used metagenomics, metatranscriptomics and mineralization assays. Mineralization of http://aem.asm.org/

29 hexadecane was significantly higher at the wellhead station at a depth of ~1200 m as compared

30 to the reference station. Community composition based on taxonomical or functional data

31 showed that the samples taken at a depth of ~1200 m were significantly more dissimilar

32 between the stations than at other depths (surface, 100 m, 750 m and >1500 m). Both Bacteria on December 24, 2018 by guest 33 and Archaea showed reduced activity at depths of ~1200 m when comparing the wellhead

34 station to the reference station, and their activity was significantly higher in surficial sediments

35 as compared to 10 cm sediments. Surficial sediments also harbored significantly different active

36 genera when compared to 5 and 10 cm sediments. For the remaining microbial parameters

37 assessed, no significant differences could be observed between the wellhead and reference

38 stations and between surface and 5-10 cm deep sediments.

113 39 Introduction

40 Following the explosion and sinking of the Deepwater Horizon (DWH) oil rig in the Gulf

41 of Mexico, an estimated 3.26 to 4.9 million barrels of light crude oil was released at a depth of

42 1544 m from April 20 to July 15, 2010, making it the largest and deepest offshore spill in United Downloaded from

43 States history (3, 22). When including gaseous hydrocarbons, like methane, the total discharge

44 was 40% higher than the abovementioned estimates (12). During the spill, a deep water oil

45 plume was detected at depths of 1000-1200 m (4, 10), but this plume was no longer detectable

46 after a few months (25), in agreement with the very high degradation rates observed in

47 laboratory incubations (10). However, most microbiological research to date has focused on the http://aem.asm.org/

48 effects of the oil spill with samples taken during the contamination event or shortly thereafter (2,

49 10, 16, 17, 20, 24, 27, 33, 34, 39), and only one study reported on the bacterial communities at

50 plume depth 1 year after the spill (41). In view of the high degradation rates observed and slow

51 mixing of deep water, it was suggested that oxygen depletion at plume depth might persist for on December 24, 2018 by guest 52 several years (1, 12, 26, 35). The cause, extent and duration of this oxygen depletion was

53 subject to debate (11, 15, 16), and it is not clear how, and if, it would impact the microbial

54 communities in the long term. Recent work also indicated that significant quantities of oil sank to

55 the sea floor (38), potentially affecting microbial communities in the sediments.

56 The microbial characterization of the water column shortly after the beginning of the spill

57 identified Oceanospirillales as a dominant group of hydrocarbon degrading organisms making

58 up as much as 90% of the of the 16S rRNA gene clone libraries (6, 10, 27, 33, 41). Shortly after

59 this, other Gammaproteobacteria affiliated with Colwellia and Cycloclasticus appeared,

60 indicative of a succession from alkane to aromatic degrading bacteria (6, 33, 39, 41). In

61 addition, other phyla of bacteria (Alteromonas, Halomonas, Pseudoalteromonas) were observed

62 in the water column (10, 39). A recent DNA-Stable Isotope Probing (SIP) study provided direct

63 evidence that most of the abovementioned taxa were in fact capable of degrading various

114 64 hydrocarbons (9). Following the spill, after the flow of hydrocarbons had been arrested,

65 methylotrophs including known methane oxidizers, became dominant in the region of the plume

66 (16). Microbial communities are at the base of several crucial biogeochemical processes in

67 marine environments, including hydrocarbon degradation. Full ecosystem recovery is intimately Downloaded from 68 linked to microbial community recovery. Microorganisms might also serve as highly sensitive

69 bioindicators (37), as they have been shown to be sensitive to very low concentration of

70 pollutants especially with regard to their transcriptome (43-45). For these reasons,

71 microorganisms could be used as indicators of pollution and ecosystem recovery through the

examination of their gene content and gene expression patterns. http://aem.asm.org/

73 Two approaches were used to determine the potential effects of the DWH blowout on

74 microbial communities more than one year after the event: 1) comparison of two water column

75 stations, one very close to the well and the second 38 km away, outside the plume area; 2)

76 depth profile of deep-sea sediment cores taken in the proximity of the Macondo well. We used a on December 24, 2018 by guest 77 shotgun metagenomic and metatranscriptomic approach and compared the microbial functions,

78 community composition and activities of the different stations with depth.

115 79 Material and methods

80 Sampling sites

81 A map of the sampling sites is provided as Fig. 1. The wellhead water column station (BM-57,

82 28.7051°, -88.4016°) was located at a distance of 3.78 km SW from the actual Deepwater Downloaded from

83 Horizon wellhead and corresponded to the plume station BM-57 used by Hazen and colleagues

84 (10). The reference water column station (A6, 28.6632°, -88.0095°) was located 37.77 km SE

85 from the Deepwater Horizon wellhead, but in the same “dome” area and was outside the plume

86 area during the spill. A series of 6 deep-sea sediment cores were collected on Nov. 16, 2011

87 during a second cruise. The cores were collected from the vicinity of the Deepwater Horizon http://aem.asm.org/

88 wellhead (around 28.715011°; -88.358703°, 3.66 km SE from the wellhead) at a depth of

89 approximately 1600 m.

91 Water and sediment sampling on December 24, 2018 by guest

92 Water samples were collected between Sept. 9-16, 2011 using either a large bailing bucket

93 (surface samples) or a CTD Niskin rosette equipped with 20 L bottles. For each depth, three

94 replicate water samples were taken. Samples were returned to the boat and immediately

95 transferred to 4 L carboys which were previously rinsed with 70% ethanol and sterile distilled

96 water. Sample filtration was started immediately after transfer using Millipore GSWP (0.22 µm

97 pore size, 47 mm diameter) filters and glass filter supports. Each 4 L water sample was filtered

98 on two filters, resulting in a total of six filters per depth. The filters were then transferred to ice

99 and then stored at -80˚C. Between samples, the glass filter supports were rinsed with 70%

100 ethanol and sterile distilled water. For the shipping of filtered samples, coolers with dry ice were

101 used and upon arrival at the lab, filters were stored at -80˚C until nucleic acid extraction was

102 performed. Water samples destined for mineralization analysis were collected from the same

116 103 carboys as used for filtration. Samples were placed in sterile 50 ml Falcon tubes and stored at

104 4˚C. Samples were shipped on ice and upon arrival at the lab were placed immediately at 4˚C.

105 Microcosms were started as soon as possible after arrival into the lab (within 24 h). Water for

106 chemical analysis was also taken and kept at 4°C until processing. Downloaded from

107 Sediments were frozen on-board the sampling vessel. Samples were shipped and

108 received frozen and stored at -20˚C until processing. Sample processing was performed at -

109 20˚C based on a protocol modified from Juck et al. (13). In brief, an approximately 5 cm strip of

110 the core sample plastic sleeve was cut and removed (from top to bottom of the core) and a

111 ‘clean’ area of the core (i.e. not contacted by the sample sleeve) was exposed using a sterile http://aem.asm.org/

112 chisel. Once this flat clean area was exposed, a sterile 1.4 cm drill bit was used to slowly drill

113 into the core sample, parallel to the core surface. The drilled core sub-sample was then

114 transferred to a sterile 50 ml Falcon tube and stored at -80˚C until extraction of nucleic acids

115 was performed. Each core was sampled at 3 different depths – ‘0 cm’ was from the surface of on December 24, 2018 by guest 116 the sediment to 1.4 cm, ‘5 cm’ was from approximately 4.3 to 5.7 cm from core surface and ’10

117 cm’ was from approximately 9.3 to 10.7 cm from the sediment surface. From the remaining core

118 samples, the material remaining at 0, 5 and 10 cm was sampled and used for hydrocarbon

119 analysis as described below.

120

121 Water microcosm mineralization assays

122 Mineralization assays using microcosms were set up using 15 ml of seawater and 14C labeled

123 (100,000 dpm) hexadecane (2.5 ppm), as sole carbon source, with no amendments added. The

124 sealed bottles containing seawater from all the different depths for both water column stations

125 and the spiked substrate were all incubated at 15˚C (the range of sample temperatures at the

126 time of collection was 4˚C (bottom samples) to 30˚C (surface samples) with orbital shaking (140

117 127 rpm) and a microcosm KOH trap (1 ml of 1.0 M KOH in a test tube). Sampling was performed at

14 128 T=3, 7, 14, 22, 28, 35, 42, 49, 56 and 63 days. The amount of radioactive CO2 produced, due

129 to the complete mineralization of the added carbon source (hexadecane) was determined by

130 scintillation counting of the KOH solution recovered from the microcosm flasks and is presented Downloaded from 14 131 as a percentage of CO2 produced from the known quantity of carbon source added at T=0.

132 During sampling of the KOH traps, atmospheric oxygen was introduced (through a 0.22 µm

133 filter) into the microcosms to ensure sufficient aeration of the samples. The 700 m depth sample

134 from the reference station was also used as a sterile abiotic control by autoclaving for 20 min

135

and cooling to room temperature before addition of the radioactive spike. http://aem.asm.org/

136

137 Hydrocarbon analyses

138 Water column samples were extracted for C10-C50 and PAH analyses using liquid-liquid

139 extraction (US EPA Method 3510 C, on December 24, 2018 by guest

140 http://www.epa.gov/osw/hazard/testmethods/sw846/pdfs/3510c.pdf). Extracts of water were

141 analyzed using high resolution gas chromatography (Agilent 6890 GC) coupled to a mass

142 selective detector (Agilent 5973N) (Willmington, DE, USA) operated in the selective ion

143 monitoring mode (SIM) using the following GC (MDN-5S column 30 m x 0.25mm id 0.25 μm film

144 thickness, Supelco Canada) conditions: cool on-column injection with oven track mode (track 3

145 °C higher that the oven temperature program) 80 °C hold 2 min ramp at 4 °C/min to 280 °C hold

146 10 min. Deep-sea sediments were processed according to King and Lee (19) and the GC-MS

147 conditions outlined for the water extracts were applied to sediment extracts.

148

149 Total DNA extraction from filters (seawater)

118 150 From each water sample (2 stations X 5 depths X 3 replicates = 30 water samples), one filter

151 was used and treated for DNA extraction, resulting in 30 DNA extracts. In the 50-ml Falcon

152 tube containing the filter, 1.7 ml of Tris-EDTA (TE) pH 8.0 buffer was added with 45 µl of 20%

153 (w/v) sodium dodecyl sulfate (SDS) and 9 µl of 20mg/ml proteinase K. The tube was incubated Downloaded from 154 with gentle inversion at 37°C for one hour. At the end of the incubation, 300 µl of 5M NaCl was

155 added in addition to 240 µl of a 10% (w/v) cetyl trimethylammonium bromide (CTAB) and 0.7M

156 NaCl solution. The tube was incubated at 65°C for 10-min. The total DNA was extracted with 1

157 volume of 24:1 chloroform/isoamyl alcohol. After centrifugation for 10 min at 3,000 x g, the

158 upper phase was transferred and mixed with one volume of 25:24:1 phenol/chloroform/isoamyl http://aem.asm.org/

159 alcohol. Following centrifugation at 16,000 x g for 10min at 4°C, the supernatant was

160 precipitated by mixing 0.6 volume of isopropanol and 1/50 volume of glycogen (5 mg/mL),

161 incubating one hour at -80°C and centrifuging at 12,000 x g for 30 min at 4°C. The DNA pellets

162 were washed using 1ml of 80% (v/v) ice-cold ethanol and dried using a SpeedVac. The DNA

163 was re-suspended in 50µl of nuclease-free water and treated with RNase If (NEB, Ipswich, MA) on December 24, 2018 by guest

164 according to the manufacturer’s instructions. After the reaction was complete and the enzyme

165 was inactivated, the DNA was purified with the QIAEX II Kit (QIAgen, Valencia, CA) and

166 quantified using the PicoGreen assay (Invitrogen).

167

168 Total RNA extraction from filters (seawater)

169 For RNA extraction, one replicate seawater sample was used (5 depths X 2 stations = 10

170 samples). All solutions were RNase free. In the 50-ml Falcon tube containing the filter, 1.6 mL of

171 freshly prepared lysozyme (10 mg/ml in TE pH 8.0) and 80 uL of 20% SDS were added and the

172 tube was then incubated at 64°C for 5 min. At the end of the incubation, 176 µl of 3M sodium

173 acetate pH 5.2 and 1.6 mL of pre-warmed acid phenol was added to the lysate incubated at

119 174 64°C for 6 min, with mixing every minute. The tube was transferred on ice for 2 min and then

175 centrifuged at 16,000 x g for 10 min at 4°C. The upper phase was transferred and mixed with

176 1.6 ml of chloroform before centrifugating at 16,000 x g for 2 min at 4°C. The upper aqueous

177 phase was transferred, mixed with 20 µl of glycogen (5 mg/ml), 160 µL of 3M sodium acetate Downloaded from 178 pH 5.2 and 4 mL of 100% ice-cold ethanol and incubated for 30 min on dry ice before

179 centrifuging at 12,000 x g for 30 min at 4°C. The pellet was washed with 1 ml of 80% ice-cold

180 ethanol and dried using a SpeedVac. The RNA was re-suspended in 400 µL of nuclease-free

181 water (Ambion, Life Technologies, Burlington, Ontario, Canada) and pooled together in the

182 same tube. The extracted total RNA was treated with Turbo DNase I (Ambion) and purified with http://aem.asm.org/

183 RNeasy MinElute Cleanup Kit (Qiagen).

184

185 DNA/RNA extraction (deep-sea sediments)

186 DNA and RNA were extracted simultaneously from 2 g of sediment using the MoBio RNA on December 24, 2018 by guest

187 PowerSoil Total RNA Isolation Kit with the RNA PowerSoil DNA Elution Accessory kit (MoBio

188 Laboratories, Carlsbad, CA).

189

190 Metagenomic sequencing

191 Each DNA library was prepared for sequencing from 50-100 ng of DNA using the Ion Xpress

192 Plus Fragment Library Kit (Life Technologies) with the Ion Xpress Barcode Adapters 1-16 (Life

193 Technologies), using the Ion Shear Plus Reagents and a Pippin Prep instrument (SAGE

194 Science, Beverly, MA) for size-selection. Barcoded libraries were pooled in an equimolar ratio

195 three by three. A total of 3.50 x 107 molecules were used in an emulsion PCR using the Ion

196 OneTouch 200 Template Kit (Life Technologies) and the OneTouch instrument (Life

120 197 Technologies). The sequencing of the pooled libraries was performed using the Personal

198 Genome Machine (PGM) system with the Ion Sequencing 200 kit and 316 chips (Life

199 Technologies). Sequencing statistics are shown in Table S1.

200 Downloaded from

201 Metatranscriptomic sequencing

202 In order to get enough RNA for library preparation, RNA samples were amplified using the

203 MessageAmp II-Bacteria Kit (Ambion) according to the manufacturer's protocol. The antisense

204 RNA (aRNA) obtained was subjected to ribosomal RNA subtraction following the procedure of http://aem.asm.org/

205 Stewart et al. (36) with the exception that the T7 promoter was coupled to the forward primer

206 instead of the reverse primer. After subtraction, a 227 bp control RNA transcribed from the

207 pSPT18 vector (positions 2867-3104 and 1-70) was added in a 1:1000 ratio (on a nanogram

208 basis) to the total rRNA-subtracted RNA. This mixture was then reverse-transcribed using the

209 SuperScript III kit (Invitrogen, Life Technologies). Illumina libraries were prepared following the on December 24, 2018 by guest

210 protocol of Meyer and Kircher (30), with tags 1 to 34. The indexed libraries were pooled in an

211 equimolar ratio and sent for eight lanes of Illumina HiSeq 2000 paired-end 2x100 bp sequencing

212 at the McGill University and Genome Quebec Innovation Centre (Montreal,

213 Canada).Sequencing statistics are shown in Table S2.

214

215 Bioinformatics

216 Metagenomic sequences were submitted to MG-RAST where they were de-replicated using the

217 method of Gomez-Alvarez et al. (8) and trimmed using the dynamic trimming method of Cox et

218 al. (5) in a way that each individual sequence would contain a maximum of 5 bases below a

219 Phred score of 15. Within MG-RAST, significant matches were defined as having 60%

121 220 sequence identity over at least 15 aa or 50 bp and with an e-value below 10-5. Metagenomic

221 data was used as relative abundance by dividing the abundance of sequences for a particular

222 organisms/gene by the total number of sequence retrieved from the sample. Metatranscriptomic

223 data resulted in 544 files (34 samples x 2 reads x 8 lanes). Data from the different lanes were Downloaded from 224 pooled together and the resulting 68 files were filtered using a custom-made Perl script, as

225 follows. Paired-end reads were processed in parallel. Reads were first trimmed at the first

226 occurrence of a low quality base (Phred score below 20) or when the adapter sequence was

227 encountered. Following this step, sequences of less than 75 bp were removed from further

228 analyses. If only one of the paired reads was filtered out, then the remaining read was also http://aem.asm.org/

229 removed. The filtered reads were then submitted to MG-RAST 3.0 (29) where mate-paired

230 reads were joined using the fastq-join utility. Mate-paired reads that did not overlap were kept

231 for downstream analyses. Within MG-RAST, significant matches were defined as having 60%

232 sequence identity over at least 15 aa or 50 bp and with an e-value below 10-5. The number of

233 sequences related to the pSPT18 vector in the filtered metatranscriptomic datasets was on December 24, 2018 by guest

234 obtained by Blast using an e-value cutoff of 10-25 and this number was used to normalize the

235 number of transcripts using the method of Moran et al. (31).

236

237 Statistical analyses

238 All statistical analyses were carried out in R (v. 2.13.2, The R foundation for statistical

239 computing, Vienna, Austria). Normal distribution of the data was tested using the “shapiro.test”

240 function. If necessary, data was then transformed using log or square root transformations.

241 Analysis of variance (ANOVA) was performed using the “aov” function while post-hoc Tukey

242 honestly significant difference tests were carried out using the “TukeyHSD” function. If these

243 transformations failed to normalize the data, a non-parametric Kruskal-Wallis test was carried

122 244 out in lieu of ANOVA (function “kruskal.test”). Correlation analyses were carried out based on

245 Spearman correlation using the “cor” function. Bray-Curtis dissimilarities were calculated using

246 the “vegdist” function of the “vegan” library.

247 Downloaded from

248 Data deposition

249 Raw sequence reads were submitted to the NCBI Sequence Read Archive (SRA) under the

250 BioProject accession PRJN0000 (pending) and the SRA project accession SRP0000 (pending).

251 Annotated metagenomes (MG) and metatranscriptomes (MT) are available in MG-RAST under http://aem.asm.org/

252 accessions 4494020.3-4494048.3 and 4494917.3 (MG, water, project 1012) 4500695.3-

253 4500711.3 (MG, sediments, project 1891), 4508873.3-4508882.3 (MT, water, project 2384) and

254 4508988.3-4509004.3 (MT, sediments, project 2866).

255 on December 24, 2018 by guest

256

123 257 Results

258 The goal of this study was to observe the effects of the DWH spill approximately one year after

259 the successful capping of the well. In order to do this, water column samples from a reference

260 station that was outside the spill area were compared to water column samples taken at similar Downloaded from

261 depths at a station that was directly in the spill area. In addition, deep-sea sediment cores were

262 taken in the direct vicinity of the well, and the surface, 5 cm and 10 cm sediment layers were

263 compared.

264 http://aem.asm.org/

265 Chemical analyses and mineralization assays

266 The chemical analyses of the water and sediments revealed very low concentration of alkanes

267 mostly in the ng per liter of water or ng per g of sediment range (Fig. 2). At these

268 concentrations, near the detection limit, variation between replicates was quite high, and the on December 24, 2018 by guest 269 only significant difference between the reference and affected water column was between the

270 surface water samples (t-test: t=3.53, P<0.05), where the reference station water column

271 samples contained significantly more alkanes (Fig. 2). The hydrocarbon measurements in the

272 deep-sea sediments were only carried out on one of the samples, so differences could not be

273 tested for significance. However, the 0 cm depth sediments showed higher concentrations of

274 alkanes (Fig. 2). For all samples (water and sediments), the majority of the alkanes detected

275 had chains longer than C20 (Fig. 2). Polycyclic aromatic hydrocarbons (PAH) and methylated

276 PAHs were below detection limits in most water samples and in all sediment samples (Fig. 2).

277 When detected in water samples, PAHs were mostly related to phenanthrene or one of its

278 methylated forms. For hexadecane mineralization assays, there was a significant difference

279 between the 1174 m sample of the wellhead water column station and the 1284 m sample of

124 280 the reference water column station (t-test: t=12.63, P<0.001), with the wellhead sample showing

281 significantly higher mineralization after 63 days of incubation (Fig. 3).

282

283 General patterns in community composition and function Downloaded from

284 Bray-Curtis dissimilarities were calculated between water column sample pairs taken at similar

285 depth based on genus relative abundance (Genus-DNA), genus-related mRNA abundance

286 (Genus-RNA), MG-RAST “functions” relative abundance (Functions-DNA) and “functions”-

287 related mRNA abundance (Functions-RNA) (Fig. 4a). For the Genus-DNA data, the Bray-Curtis http://aem.asm.org/

288 dissimilarities were significantly higher for the 1174-1284m sample pairs than for any other

289 sample pairs from comparable depths (one-way ANOVA: F=11.759, P<0.001) (Fig. 4a),

290 suggesting more dissimilar communities between the stations in water samples taken at this

291 depth. In contrast, for the Functions-DNA, no significant differences between the average

292 dissimilarity for water samples taken at similar depths for the two stations were detected, on December 24, 2018 by guest

293 indicating that similarity in the functional potential did not vary with depth. For the RNA data,

294 only one water sample per depth was analyzed, so significance could not be tested. However,

295 the Bray-Curtis dissimilarities were higher for the 1174-1284 m sample pair for both Genus and

296 Functions (Fig. 4a), indicating that, at this depth, there are relatively more differences in active

297 organisms and gene expression between the two water column stations.

298 For the deep-sea sediments, we compared the different depths among and between

299 each other. The only significant differences observed were for the Genus-RNA datasets, where

300 the Bray-Curtis dissimilarities within the surface layers of the six sediment cores were higher

301 than the distance within the 5 or 10 cm deep layers (one-way ANOVA: F=12.48, P<0.001),

302 suggesting a higher heterogeneity in the active community composition in the surface

303 sediments. Furthermore, when comparing the Bray-Curtis dissimilarities between the different

125 304 sediment layers, the dissimilarities were significantly higher when the surface layer was involved

305 (one-way ANOVA: F=12.25, P<0.001) (Fig. 4b), indicating that the active microbial community

306 at the surface of the sediment is different from the one in the deeper layers.

307 Downloaded from

308 Microbial community composition and activity

309 As expected, Cyanobacteria were dominant in surface water samples from the wellhead and the

310 reference column stations, while deeper water samples were dominated by Proteobacteria,

311 mostly from the Gammaproteobacteria subclass (Fig. 5). The Thaumarchaeota showed a http://aem.asm.org/

312 particular pattern, having very high relative abundance at a depth of 1174 m (nearly 40%) for

313 the wellhead water column station and at a depth of 700m (over 30%) for the reference water

314 column station (Fig. 5). When comparing the two water column stations, Thaumarchaeota were

315 significantly more abundant at the wellhead water column station for the 1174-1284 m depths (t-

316 test: t=5.3021, P<0.05). The majority of the Thaumarchaeota sequences were related to the on December 24, 2018 by guest

317 Nitrosopumilus genus. At the phylum level, the deep-sea sediments appeared relatively

318 homogenous, with small increases in Deltaproteobacteria and small decreases in

319 Alphaproteobacteria with increasing depth below the seafloor (Fig. 5).

320 The Oceanospirillales have been previously reported as being dominant in the water

321 column during the DWH oil spill. Oceanospirillales were not very abundant in our samples,

322 forming less than 3% of all DNA reads in the water column samples and less than 2% of all

323 DNA reads in the deep-sea sediment samples. Oceanospirillalles were significantly more

324 abundant at the reference water column station at depths of 700-850m, but that did not strongly

325 affect their activity . However, the activity in the water column at depths of 1174-1284m was two

326 orders of magnitude lower at the wellhead water column station. In the deep-sea sediments,

327 Oceanospirillales were significantly more active (F=4.31, P<0.05) and abundant (F=7.93,

126 328 P<0.05) in the surface sediments as compared to the 10 cm deep sediments, with the 5 cm

329 deep sediments showing intermediate values (DNA average relative abundance of 1.70% in the

330 surface sediments, 0.62% at 5 cm depth and 0.37% at 10 cm depth). The obligate

331 hydrocarbonoclastic bacteria (OHCB), Alcanivorax and Marinobacter did not show any Downloaded from 332 significant differences between the two water column stations at the DNA level (DNA average

333 relative abundance of 0.16% and 0.24%, respectively), but were less active at plume depth

334 (1174-1284m) in the wellhead water column station as compared to the reference water column

335 station. For the sediments, no significant differences were observed at the DNA level (DNA

336 average relative abundance of 0.28% for Alcanivorax and 0.022% for Marinobacter), but at the http://aem.asm.org/

337 RNA level, both Marinobacter (F=7.69, P<0.05) and Alcanivorax (F=9.84, P<0.01) showed

338 higher activity in the surface sediments. Colwellia did not show any significant differences

339 between the two water column stations or the different sediment depths at the DNA level (DNA

340 average relative abundance of 0.087% for the water column samples and of 0.13% in the

341 sediments). Colwellia was significantly more active in the surface sediments as compared to the on December 24, 2018 by guest

342 5 and 10 cm deep sediments (F=9.13, P<0.05) and showed maximum activity in the deepest

343 water column samples for both stations. Methanotroph DNA and RNA were detected at all

344 depths for both water column stations and for the sediments with a dominance of Type I

345 methanotrophs, from the Methylococcaceae family (DNA average relative abundance of 0.17%

346 for Methylococcaceae and 0.0078% for Methylocystaceae in the water column samples, and

347 0.65% for Methylococcaceae and 0.00033% for Methylocystaceae in the sediment samples).

348 The only statistically significant effect was a higher activity of Methylococcaceae in the surface

349 sediments (F=7.83, P<0.05). Methanotrophs were also one order of magnitude more active at

350 the wellhead water column station when comparing the deepest water samples to each other

351 (1574-2174m).

127 352 The activity of Archaea, as measured by the abundance of related mRNA, was slightly

353 higher in the wellhead water column samples with the exception of the 1174-1284 m water

354 samples, where Archaea were less active at the wellhead water column station. A similar

355 pattern emerged for bacterial activity, with the exception that the differences were larger and Downloaded from 356 Bacteria were more active at the reference water column station for the surface samples. For

357 the sediments, Archaea were significantly more active in the surface layer (F=40.29, P<0.001),

358 while Bacteria were significantly less active in the surface layer (F=9.94, P<0.01). These

359 patterns at the domain level were not significant in the metagenomic datasets for both water

360

column and sediment samples. http://aem.asm.org/

361

362 Microbial hydrocarbon degradation related functions

363 For both sediments and water samples, no significant differences were found for the relative

364 abundance of aerobic (alkane monooxygenase and ring-opening dioxygenase) and anaerobic on December 24, 2018 by guest

365 (acetyl-CoA acetyltransferase and benzoyl-CoA reductase) key hydrocarbon degradation genes

366 based on the DNA dataset. For gene expression based on RNA sequencing, some interesting

367 trends emerged (Fig. 6). For instance, the alkane monooxygenase gene was not expressed in

368 the surface waters (0-100 m) of the reference water column station, while it was not strongly

369 expressed in the deeper waters of the wellhead water column station (1174-1574 m) (Fig. 6a).

370 In contrast the ring-opening dioxygenases were more expressed at the wellhead water column

371 station for depths greater than 700 m, sometimes by several orders of magnitude (Fig. 6a). The

372 two anaerobic genes were less expressed at the wellhead water column station at depths of

373 1174-1284 m, but the benzoyl-CoA reductase gene was more expressed at the wellhead water

374 column station for depths of 750-800 m (Fig. 6a). For the sediments, the only significant

128 375 difference was in the expression of alkane monooxygenase, that was significantly more

376 expressed in the top layer of the sediments (F=9.71, P<0.01) (Fig. 6b).

377

378 Starvation-related functions Downloaded from

379 The expression of gene categories related to nutrient deficiency (e.g. siderophore production,

380 carbon starvation, nutrient transporters) was examined. When comparing the two water column

381 stations, siderophore genes were more expressed at the wellhead water column station at

382 1174-1284 m, while they were more expressed at the reference water column station at the http://aem.asm.org/

383 surface (Fig. 7a). Carbon starvation genes were not expressed at the surface and at depths of

384 100 m and 2174 m for the reference water column station, while these genes were expressed

385 throughout the depth profile for the wellhead water column station, except for the surface

386 samples (Fig. 7a). The highest expression of carbon starvation related genes was observed at a

387 depth of 850 m for the wellhead water column station (Fig. 7a). Ammonium transporters were on December 24, 2018 by guest

388 expressed similarly at most of the depths of the two water column stations, with the exception of

389 the surface and the 1174-1284 m depths, where the expression was higher at the reference

390 water column station (Fig. 7a). In contrast, the expression of genes related to phosphate

391 starvation was higher for the reference water column station at almost all depths, except the

392 700-850 m depth (Fig. 7a). In the sediments, most of the gene categories were less expressed

393 in the surface of the sediment cores, except for the phosphate starvation genes that were

394 significantly less expressed in the 10 cm deep sediments (F=4.59, P<0.05) (Fig. 7b). The only

395 other significant trend was for ammonium transporters that were significantly less expressed in

396 the surface sediments (F=9.17, P<0.05) (Fig. 7b).

129 397 Discussion

398 In the present study, we looked simultaneously at the metatranscriptome and metagenome of

399 water samples taken at an offshore station strongly affected by the DWH spill and at another

400 offshore station that was unaffected by the spill, approximately one year after the capping of the Downloaded from

401 well. We also analyzed deep sea sediments from the surface to 10 cm deep adjacent to the

402 well.

403

404 Water http://aem.asm.org/

405 A large part of our dataset showed no significant differences between the water of the

406 reference and wellhead stations. No trace of the massive amount of oil released during the spill

407 could be found, with most hydrocarbon concentrations being near or below the detection limit

408 and not significantly different between the water of the wellhead and reference stations. on December 24, 2018 by guest 409 Oceanospirillales relative abundance was also not different between the water column stations,

410 after being reported as largely dominant during the spill (10, 27). No differences were observed

411 between the water of the wellhead and reference stations for the relative abundance and

412 expression of hydrocarbon degradation genes and the presence of obligate hydrocarbonclastic

413 bacteria (OHCB). Kessler et al. (16) hypothesized that methanotrophs were no longer active in

414 September 2010 even though they were detected in high numbers, because methane

415 concentrations and oxidation rates became very low. Here, we found methanotrophs to be

416 present and active at all depths for the two water column stations, as previously reported in non-

417 plume samples (34) and in post-spill samples (41). Since methanotrophs can only use methane

418 or methanol as their carbon source, their presence and activity at all depths in the water column

419 is indirectly indicative of the presence of either methane or methanol. The amount of gas

420 emitted during the spill comprised 40% (500,000 t) of the total hydrocarbon discharge (12), but

130 421 was reported to be degraded very rapidly (16). Another study recently suggested that the genus

422 Colwellia was responsible for the majority of ethane and propane oxidation during the

423 Deepwater Horizon spill (33). In our study, this genus had very low relative abundance, but was

424 active in several water column and sediment samples, with relatively more transcripts in the Downloaded from 425 deepest samples and in the sediments. Similarly, several hydrocarbonoclastic bacteria were

426 present in GOM water 1 month before (18) and up to 1 year after (41) the DWH accident.

427 These, and our results suggest that in the water of the Gulf of Mexico, although hydrocarbon

428 concentrations are typically very low, there are permanent microbiological activities related to

429 the degradation of hydrocarbons. These activities are probably carried out by a minority of the http://aem.asm.org/

430 microbial community (as hydrocarbon-degraders had very low relative abundance), but upon

431 feeding with fresh hydrocarbon substrate (as in the microcosm experiments), this minority could

432 be rapidly stimulated. This continuous background hydrocarbon degradation activity probably

433 explains the very rapid disappearance of the hydrocarbon released during the DWH spill (10).

434 This continuous hydrocarbon degradation activity also results in the maintenance of the genetic on December 24, 2018 by guest

435 potential for the degradation of hydrocarbons in the indigenous microbial communities. The

436 natural seepage of hydrocarbons from the GOM seafloor was estimated to amount 4–10 × 1010

437 g per year (32), which results in constant exposure of the microbial community to hydrocarbons,

438 and since this is essentially a continuous process, residual hydrocarbons may remain very low

439 even though hydrocarbon degradation activity is quite high.

440 Several intriguing differences were observed between the water of the reference and

441 wellhead stations at depths where a dissolved hydrocarbon plume was detected during the spill

442 (~1200 m). For instance, there were significantly higher mineralization rates in the water of the

443 wellhead column station, indicating a residual higher potential for alkane degradation. However,

444 no significant differences were observed for hydrocarbon-degrading genes in the metagenomic

445 datasets, suggesting that the potential for hydrocarbon degradation is not necessarily related to

446 the presence of specific functional genes. Rather, this potential might be related to the identity

131 447 of the microorganisms that harbor these genes, as the water microbial communities were

448 significantly more dissimilar at depths around 1200 m, suggesting that unique microbial

449 communities were present in the water of the wellhead station at depths where the plume was

450 found. Indeed, some hydrocarbon-degrading organisms have higher metabolic rates and their Downloaded from 451 presence in the water samples could have increased mineralization rates in the microcosm

452 experiments. We can only hypothesize on the reasons behind these differences: presence of

453 natural seeps, low mixing rates in the deeper water or some kind of persistent anomaly at plume

454 depth.

455 Another difference between the wellhead and the reference water column stations was http://aem.asm.org/

456 that Thaumarchaeota were significantly more abundant at the wellhead at ~1200 m. All

457 organisms of this lineage thus far identified are chemolithoautotrophic ammonia-oxidizers and

458 Thaumarchaeota sequences made up more than 40% of the metagenomic reads at the

459 wellhead water column station, from which the majority were related to Nitrosopumilus. Archaea on December 24, 2018 by guest 460 were previously reported to dominate the mesopelagic zone (14). This high abundance was,

461 however, very surprising in view of the sensitivity of Nitrosopumilus and other ammonia-

462 oxidizers to Macondo crude oil (37). Nitrosopumilus was even proposed as a bio-indicator to

463 map future spills (37). In contrast, Rivers et al. (34) reported that Nitrosopumilus were similarly

464 active in plume and non-plume samples during the DWH spill. Nitrosopumilus maritimus was

465 reported to be dominant in the suboxic zone of the Baltic Sea (21), and the lower oxygen

466 concentrations observed in some areas after the spill (up to 30-50% oxygen depletion (12))

467 could have favored Nitrosopumilus. Although some studies predicted this localized oxygen

468 depletion to persist for months to years because of slow water mixing rates at depth (1, 12, 16,

469 26, 35), the causes, extent and duration of this depletion have been debated (11, 15). Modeling

470 efforts suggested that physical dynamics would result in the absence of an extensive and

471 persistent oxygen depletion in the deep plume horizon (40), and recent studies of the microbial

132 472 dynamics suggested that the oxygen anomaly observed at plume depth after the spill could be

473 due to consumption of organic matter from dead organisms in the plume (6).

474 Microbial activities in the plume could also have resulted in localized nutrient depletion.

475 Nutrient limitation was previously reported during surface oil degradation at an offshore station Downloaded from

476 during the DWH oil spill, as alkaline phosphatase activity increased, indicative of phosphate

477 limitation. Bacterial respiration also increased when hydrocarbons were present but without a

478 concomitant increase in bacterial biomass, which only increased upon nutrient addition (7).

479 Nitrosopumilus is especially efficient at low nutrient concentrations, which might explain its

480 predominance in the plume depth wellhead samples. Genes involved in iron limitation were http://aem.asm.org/

481 overexpressed at plume depth (1174 m) at the wellhead station, further suggesting nutrient

482 limitation at these depths. Another study revealed that the draft genome sequence of a

483 dominant Oceanospirillales contained many genes related to the uptake of various nutrients,

484 including ammonium, phosphate and iron, all of which were also found in the metagenomes and on December 24, 2018 by guest 485 expressed in the plume metatranscriptome (27). Some of the differences observed between the

486 reference and wellhead stations at plume depth could have been related to a persistent nutrient

487 depletion. However, during the spill, nitrogen and phosphorus did not appear to be limiting

488 based on measured values at plume depth (34) and other factors might explain the differences

489 observed in microbial communities between the affected and unaffected water columns at the

490 depths where the oil plume was found.

491

492 Sediments

493 The sedimentation rate in the Gulf of Mexico was reported to be 0.09 cm/year at a

494 station located 70.91 km away from the wellhead station, at similar depths (1849 m) (42). Other

495 stations in the same area and at similar depths also showed very similar sedimentation rates.

133 496 Bioturbation was reported to be responsible for the active mixing that occurred in the top 2-3 cm

497 and macrofaunal density was significantly correlated with organic carbon inventory (42).This

498 probably resulted in an increased heterogeneity for the surficial sediments, supported by the

499 higher dissimilarity observed for the genus activity between the six replicate surface sediments Downloaded from 500 as compared to 5 and 10 cm deep sediments.

501 In both the DNA and RNA datasets, the microbial communities in the surface layer

502 sediments were very often significantly different from the 10 cm layer and sometimes from the 5

503 cm layer. This is consistent with the very sharp redox gradient in deep-sea sediments, with

504 microbial communities in surface sediments being under aerobic conditions and those in deeper http://aem.asm.org/

505 sediments under anaerobic conditions. Surficial sediments in our study were co-dominated by

506 the generally aerobic Alphaproteobacteria and Gammaprotebacteria, consistent with previous

507 reports on the bacterial communities of sediments taken less than 6 km away from the wellhead

508 one year after the DWH spill (23). In contrast, directly after the spill, the relative abundance of on December 24, 2018 by guest 509 Deltaproteobacteria anaerobic hydrocarbon degraders and associated functional genes was

510 higher in surficial sediments closest to the blowout site (17). In our case, Deltaproteobacteria

511 were not very abundant in the surface layers, but increased in abundance with depth, in line

512 with the expected decrease in sediment oxygen concentration with depth. Similarly, the

513 expression of genes related to oxygen-dependent enzymes, like the alkane monooxygenase,

514 were significantly more expressed at the surface of the sediments as compared to deeper

515 sediment. Hydrocarbon degradation genes were reported to be more abundant in highly

516 contaminated surficial sediments shortly after the spill (28).

517 The alkane concentrations in our surficial sediments were similar to the ones measured

518 by Mason et al. (28) in sediments taken less than 5 km from the wellhead in October 2010

519 (4,619 ng g-1 here, vs. average of 7,304 ng g-1 for Mason et al.), suggesting very slow

520 degradation, if any, in the >1 year between the samplings. In contrast, PAHs were undetected in

134 521 our surficial sediments as compared to an average of 1,895 ng g-1 with only one sample

522 showing concentrations below the detection limit in Mason et al. (28). Taken together, these

523 results suggest that PAHs were probably degraded more rapidly in the surficial sediments near

524 the DWH wellhead than the alkanes, even though surficial sediments showed higher Downloaded from 525 mineralization rates for dodecane (C12) than for phenanthrene or toluene at 5°C (28). The bulk

526 of the alkanes detected in our surficial sediments had chains longer than C23 (82%; 3,765 ng g-

527 1 out of 4,619 ng g-1), which are generally more recalcitrant to microbial degradation.

528 http://aem.asm.org/ 529 Conclusions

530 Most of the data indicated no significant differences between the wellhead water column station

531 and the reference station. This included some hydrocarbonoclastic activity, probably related to

532 the presence of hydrocarbons from natural seeps. Hydrocarbon concentrations in the water

533 column were very low, in the ng/L range, indicating that very little, if any, hydrocarbon from the on December 24, 2018 by guest

534 DWH spill remained in the water column on the date and at the location sampled, in agreement

535 with studies that reported a very rapid degradation of hydrocarbons from the spill (10). However,

536 a few microbial indicators showed significant differences between the reference and wellhead

537 stations at depth where the hydrocarbon plume was detected (~1200 m), most likely related to a

538 persistent nutrient or oxygen limitation, or a legacy effect, rather than to the presence of residual

539 hydrocarbons from the spill. As for the sediments, several significant differences were observed

540 between surficial and deeper sediments, probably related to differences in geochemical

541 conditions (mainly oxygen availability). Surficial sediments collected near the wellhead

542 contained alkane concentrations similar to those measured a few months after the spill,

543 probably because of the higher recalcitrance of the long-chain alkanes detected.

544

135 545 Acknowledgments

546 Suzanne Labelle, Claude Masson and Danielle Ouellette from the NRC and Thomas King from

547 DFO are thanked for their excellent technical support in sample preparation and analyses. We

548 are thankful to Arden Ahnell and Marie BenKinney for insightful comments and support Downloaded from

549 throughout this study. This study was supported by British Petroleum, but they neither

550 participated in the preparation of this manuscript nor in the analysis or interpretation of the

551 results.

552 http://aem.asm.org/ on December 24, 2018 by guest

136 553 References

554 1. Adcroft, A., R. Hallberg, J. P. Dunne, B. L. Samuels, J. A. Galt, C. H. Barker, and D. Payton. 2010. 555 Simulations of underwater plumes of dissolved oil in the Gulf of Mexico. Geophys. Res. Lett. 556 37:L18605. 557 2. Bælum, J., S. Borglin, R. Chakraborty, J. L. Fortney, R. Lamendella, O. U. Mason, M. Auer, M.

558 Zemla, M. Bill, M. E. Conrad, S. A. Malfatti, S. G. Tringe, H.-Y. Holman, T. C. Hazen, and J. K. Downloaded from 559 Jansson. 2012. Deep-sea bacteria enriched by oil and dispersant from the Deepwater Horizon 560 spill. Environ. Microbiol. 14:2405-2416. 561 3. BP, accessed July 10, 2014. Containing the leak. British Petroleum. [Online.] 562 4. Camilli, R., C. M. Reddy, D. R. Yoerger, B. A. S. Van Mooy, M. V. Jakuba, J. C. Kinsey, C. P. 563 McIntyre, S. P. Sylva, and J. V. Maloney. 2010. Tracking Hydrocarbon Plume Transport and 564 Biodegradation at Deepwater Horizon. Science 330:201-204. 565 5. Cox, M., D. Peterson, and P. Biggs. 2011. SolexaQA: At-a-glance quality assessment of Illumina 566 second-generation sequencing data. BMC Bioinf. 11:485.

567 6. Dubinsky, E. A., M. E. Conrad, R. Chakraborty, M. Bill, S. E. Borglin, J. T. Hollibaugh, O. U. http://aem.asm.org/ 568 Mason, Y. M. Piceno, F. C. Reid, and W. T. Stringfellow. 2013. Succession of hydrocarbon- 569 degrading bacteria in the aftermath of the Deepwater Horizon oil spill in the Gulf of Mexico. 570 Environ. Sci. Technol. 47:10860-10867. 571 7. Edwards, B. R., C. M. Reddy, R. Camilli, C. A. Carmichael, K. Longnecker, and B. A. S. Van Mooy. 572 2011. Rapid microbial respiration of oil from theDeepwater Horizonspill in offshore surface 573 waters of the Gulf of Mexico. Environ. Res. Lett. 6:035301. 574 8. Gomez-Alvarez, V., T. K. Teal, and T. M. Schmidt. 2009. Systematic artifacts in metagenomes 575 from complex microbial communities. ISME J. 3:1314–1317.

576 9. Gutierrez, T., D. R. Singleton, D. Berry, T. Yang, M. D. Aitken, and A. Teske. 2013. Hydrocarbon- on December 24, 2018 by guest 577 degrading bacteria enriched by the Deepwater Horizon oil spill identified by cultivation and 578 DNA-SIP. ISME J 7:2091-2104. 579 10. Hazen, T. C., E. A. Dubinsky, T. Z. DeSantis, G. L. Andersen, Y. M. Piceno, N. Singh, J. K. Jansson, 580 A. Probst, S. E. Borglin, J. L. Fortney, W. T. Stringfellow, M. Bill, M. E. Conrad, L. M. Tom, K. L. 581 Chavarria, T. R. Alusi, R. Lamendella, D. C. Joyner, C. Spier, J. Baelum, M. Auer, M. L. Zemla, R. 582 Chakraborty, E. L. Sonnenthal, P. D'Haeseleer, H. Y. N. Holman, S. Osman, Z. M. Lu, J. D. Van 583 Nostrand, Y. Deng, J. Z. Zhou, and O. U. Mason. 2010. Deep-Sea Oil Plume Enriches Indigenous 584 Oil-Degrading Bacteria. Science 330:204-208. 585 11. Joye, S. B., I. Leifer, I. R. MacDonald, J. P. Chanton, C. D. Meile, A. P. Teske, J. E. Kostka, L. 586 Chistoserdova, R. Coffin, D. Hollander, M. Kastner, J. P. Montoya, G. Rehder, E. Solomon, T. 587 Treude, and T. A. Villareal. 2011. Comment on “A Persistent Oxygen Anomaly Reveals the Fate 588 of Spilled Methane in the Deep Gulf of Mexico”. Science 332:1033. 589 12. Joye, S. B., I. R. MacDonald, I. Leifer, and V. Asper. 2011. Magnitude and oxidation potential of 590 hydrocarbon gases released from the BP oil well blowout. Nat. Geosci. 4:160-164. 591 13. Juck, D. F., G. Whissell, B. Steven, W. Pollard, C. P. McKay, C. W. Greer, and L. G. Whyte. 2005. 592 Utilization of fluorescent microspheres and a green fluorescent protein-marked strain for 593 assessment of microbiological contamination of permafrost and ground ice core samples from 594 the Canadian High Arctic. Appl. Environ. Microbiol. 71:1035–1041. 595 14. Karner, M. B., E. F. DeLong, and D. M. Karl. 2001. Archaeal dominance in the mesopelagic zone 596 of thePacific Ocean. Nature 409:507-510.

137 597 15. Kessler, J. D., D. L. Valentine, M. C. Redmond, and M. Du. 2011. Response to Comment on “A 598 Persistent Oxygen Anomaly Reveals the Fate of Spilled Methane in the Deep Gulf of Mexico”. 599 Science 332:1033. 600 16. Kessler, J. D., D. L. Valentine, M. C. Redmond, M. Du, E. W. Chan, S. D. Mendes, E. W. Quiroz, 601 C. J. Villanueva, S. S. Shusta, L. M. Werra, S. A. Yvon-Lewis, and T. C. Weber. 2011. A Persistent 602 Oxygen Anomaly Reveals the Fate of Spilled Methane in the Deep Gulf of Mexico. Science 603 331:312-315. 604 17. Kimes, N. E., A. V. Callaghan, D. F. Aktas, W. L. Smith, J. Sunner, B. Golding, M. Drozdowska, T. Downloaded from 605 C. Hazen, J. M. Suflita, and P. J. Morris. 2013. Metagenomic analysis and metabolite profiling of 606 deep-sea sediments from the Gulf of Mexico following the Deepwater Horizon oil spill. Front 607 Microbiol 4:50. 608 18. King, G. M., C. Smith, B. Tolar, and J. T. Hollibaugh. 2013. Analysis of composition and structure 609 of coastal to mesopelagic bacterioplankton communities in the northern Gulf of Mexico. Front 610 Microbiol 3. 611 19. King, T. L., and L. K. 2004. Presented at the Proceedings of the 27th Arctic and Marine Oilspill 612 Program (AMOP), Edmonton, Alberta, Canada. http://aem.asm.org/ 613 20. Kostka, J. E., O. Prakash, W. A. Overholt, S. J. Green, G. Freyer, A. Canion, J. Delgardio, N. 614 Norton, T. C. Hazen, and M. Huettel. 2011. Hydrocarbon-Degrading Bacteria and the Bacterial 615 Community Response in Gulf of Mexico Beach Sands Impacted by the Deepwater Horizon Oil 616 Spill. Appl. Environ. Microbiol. 77:7962-7974. 617 21. Labrenz, M., E. Sintes, F. Toetzke, A. Zumsteg, G. J. Herndl, M. Seidler, and K. Jurgens. 2010. 618 Relevance of a crenarchaeotal subcluster related to Candidatus Nitrosopumilus maritimus to 619 ammonia oxidation in the suboxic zone of the central Baltic Sea. ISME J 4:1496-508. 620 22. Lehr, B., S. Bristol, and A. Possolo. 2010. Oil Budget Calculator, Deepwater Horizon. The Federal 621 Interagency Solutions Group, Oil Budget Calculator Science and Engineering Team, Washington, 622 DC. on December 24, 2018 by guest 623 23. Liu, Z., and J. Liu. 2013. Evaluating bacterial community structures in oil collected from the sea 624 surface and sediment in the northern Gulf of Mexico after the deepwater horizon oil spill. 625 MicrobiologyOpen 2:492-504. 626 24. Lu, Z., Y. Deng, J. D. Van Nostrand, Z. He, J. Voordeckers, A. Zhou, Y.-J. Lee, O. U. Mason, E. A. 627 Dubinsky, K. L. Chavarria, L. M. Tom, J. L. Fortney, R. Lamendella, J. K. Jansson, P. D'Haeseleer, 628 T. C. Hazen, and J. Zhou. 2012. Microbial gene functions enriched in the Deepwater Horizon 629 deep-sea oil plume. ISME J 6:451-460. 630 25. Mascarelli, A. 2010. Deepwater Horizon: After the oil. Nature 467:22-24. 631 26. Mascarelli, A. 2010. Muddying the waters on Gulf oxygen data. Nature doi: 632 10.1038/news.2010.378. 633 27. Mason, O. U., T. C. Hazen, S. Borglin, P. S. G. Chain, E. A. Dubinsky, J. L. Fortney, J. Han, H.-Y. N. 634 Holman, J. Hultman, R. Lamendella, R. Mackelprang, S. Malfatti, L. M. Tom, S. G. Tringe, T. 635 Woyke, J. Zhou, E. M. Rubin, and J. K. Jansson. 2012. Metagenome, metatranscriptome and 636 single-cell sequencing reveal microbial response to Deepwater Horizon oil spill. ISME J 6:1715- 637 1727. 638 28. Mason, O. U., N. M. Scott, A. Gonzalez, A. Robbins-Pianka, J. Balum, J. Kimbrel, N. J. Bouskill, 639 E. Prestat, S. Borglin, D. C. Joyner, J. L. Fortney, D. Jurelevicius, W. T. Stringfellow, L. Alvarez- 640 Cohen, T. C. Hazen, R. Knight, J. A. Gilbert, and J. K. Jansson. 2014. Metagenomics reveals 641 sediment microbial community response to Deepwater Horizon oil spill. ISME J 8:1464-1475. 642 29. Meyer, F., D. Paarmann, M. D'Souza, R. Olson, E. M. Glass, M. Kubal, T. Paczian, A. Rodriguez, 643 R. Stevens, A. Wilke, J. Wilkening, and R. A. Edwards. 2008. The metagenomics RAST server - a

138 644 public resource for the automatic phylogenetic and functional analysis of metagenomes. BMC 645 Bioinformatics 9:386. 646 30. Meyer, M., and M. Kircher. 2010. Illumina Sequencing Library Preparation for Highly 647 Multiplexed Target Capture and Sequencing. Cold Spring Harb Protoc 648 doi:10.1101/pdb.prot5448. 649 31. Moran, M. A., B. Satinsky, S. M. Gifford, H. Luo, A. Rivers, L. K. Chan, J. Meng, B. P. Durham, C. 650 Shen, V. A. Varaljay, C. B. Smith, P. L. Yager, and B. M. Hopkinson. 2013. Sizing up 651 metatranscriptomics. ISME J 7:237-43. Downloaded from 652 32. National_Research_Council. 2003. Oil in the Sea III: Inputs, Fates, and Effects. National 653 Academies Press, Washington, DC. 654 33. Redmond, M. C., and D. L. Valentine. 2012. Natural gas and temperature structured a microbial 655 community response to the Deepwater Horizon oil spill. P. Natl. Acad. Sci. USA 109:20292- 656 20297. 657 34. Rivers, A. R., S. Sharma, S. G. Tringe, J. Martin, S. B. Joye, and M. A. Moran. 2013. 658 Transcriptional response of bathypelagic marine bacterioplankton to the Deepwater Horizon oil 659 spill. ISME J 7:2315-2329. http://aem.asm.org/ 660 35. Schrope, M. 2010. Oil cruise finds deep-sea plume. Nature 465:274-275. 661 36. Stewart, F. J., E. A. Ottesen, and E. F. DeLong. 2010. Development and quantitative analyses of 662 a universal rRNA-subtraction protocol for microbial metatranscriptomics. ISME J. 4:896-907. 663 37. Urakawa, H., J. C. Garcia, P. D. Barreto, G. A. Molina, and J. C. Barreto. 2012. A sensitive crude 664 oil bioassay indicates that oil spills potentially induce a change of major nitrifying prokaryotes 665 from the archaea to the bacteria. Environ. Pollut. 164:42-5. 666 38. Valentine, D. L., G. B. Fisher, S. C. Bagby, R. K. Nelson, C. M. Reddy, S. P. Sylva, and M. A. Woo. 667 2014. Fallout plume of submerged oil from Deepwater Horizon. Proc Natl Acad Sci USA 668 111:15906-15911. 669 39. Valentine, D. L., J. D. Kessler, M. C. Redmond, S. D. Mendes, M. B. Heintz, C. Farwell, L. Hu, F. on December 24, 2018 by guest 670 S. Kinnaman, S. Yvon-Lewis, M. R. Du, E. W. Chan, F. G. Tigreros, and C. J. Villanueva. 2010. 671 Propane Respiration Jump-Starts Microbial Response to a Deep Oil Spill. Science 330:208-211. 672 40. Valentine, D. L., I. Mezic, S. Macesic, N. Crnjaric-Zic, S. Ivic, P. J. Hogan, V. A. Fonoberov, and S. 673 Loire. 2012. Dynamic autoinoculation and the microbial ecology of a deep water hydrocarbon 674 irruption. Proc Natl Acad Sci U S A 109:20286-91. .Ambrosio, S. B. Joye, R. Highsmith, and A. Teske. 2014׳Yang, T., L. M. Nigro, T. Gutierrez, L. D .41 675 676 Pulsed blooms and persistent oil-degrading bacterial populations in the water column during 677 and after the Deepwater Horizon blowout. Deep Sea Res. Part II In press:DOI: 678 10.1016/j.dsr2.2014.01.014. 679 42. Yeager, K. M., P. H. Santschi, and G. T. Rowe. 2004. Sediment accumulation and radionuclide 680 inventories (239,240Pu, 210Pb and 234Th) in the northern Gulf of Mexico, as influenced by 681 organic matter and macrofaunal density. Mar. Chem. 91:1-14. 682 43. Yergeau, E., J. R. Lawrence, D. R. Korber, M. J. Waiser, and C. W. Greer. 2010. Meta- 683 transcriptomic analysis of the response of river biofilms to pharmaceutical products using 684 anonymous DNA microarrays. Appl. Environ. Microbiol. 76:5432-5439. 685 44. Yergeau, E., J. R. Lawrence, S. Sanschagrin, J. L. Roy, G. D. W. Swerhone, D. R. Korber, and C. 686 W. Greer. 2013. Aerobic biofilms grown from Athabasca watershed sediments are inhibited by 687 increasing bituminous compounds concentrations. Appl. Environ. Microbiol. 79:7398-7412. 688 45. Yergeau, E., S. Sanschagrin, M. J. Waiser, J. R. Lawrence, and C. W. Greer. 2012. Sub-inhibitory 689 concentrations of different pharmaceutical products affect the meta-transcriptome of river 690 biofilm communities cultivated in rotating annular reactors. Environ. Microbiol. Rep. 4:350-359.

139 691 Figures legends

692 Figure 1. Map of the sampling sites.

693 Figure 2. C10-C16 and C17-C35 alkane and polycyclic aromatic hydrocarbons (PAH) and

694 methylated PAHs concentrations for the wellhead and reference station water column samples Downloaded from

695 and for the deep-sea sediments . *P<0.05, **P<0.01, ***P<0.001. Note that PAH and

696 methylated PAH were not detected in any of the deep-sea sediments samples.

697 Figure 3. Hexadecane mineralization for the wellhead and reference station water column

698 samples. At day 63, different letters indicate significant differences (P<0.05) in Tukey HSD http://aem.asm.org/ 699 post-hoc test.

700 Figure 4. Average Bray-Curtis dissimilarities based on genus-level community composition or

701 expression patterns or MG-RAST function-level community composition or expression patterns

702 calculated for (a) water samples taken at similar depth at the reference and wellhead stations

703 and (b) replicate sediments taken at similar depth and sediments taken at different depths. on December 24, 2018 by guest

704 Figure 5. Community composition at the phylum/class level for the different stations and

705 sediment sampled based on the taxonomic affiliation of bacterial metagenomic (DNA) reads in

706 MG-RAST.

707 Figure 6. Normalized expression of key aerobic and anaerobic hydrocarbon-degradation genes

708 for (a) the different stations and (b) sediment sampled based on the taxonomic affiliation of

709 bacterial metagenomic (DNA) and metatranscriptomic (RNA) reads in MG-RAST. In (a): no

710 replicates were available, significance could not be tested, in (b): different letters indicate

711 significant differences (P<0.05) in Tukey HSD post-hoc test.

712 Figure 7. Normalized expression of genes related to nutrient depletion for (a) the different

713 stations and (b) sediment sampled based on the taxonomic affiliation of bacterial metagenomic

140 714 (DNA) and metatranscriptomic (RNA) reads in MG-RAST. In (a): no replicates were available,

715 significance could not be tested, in (b): different letters indicate significant differences (P<0.05)

716 in Tukey HSD post-hoc test.

717 Downloaded from http://aem.asm.org/ on December 24, 2018 by guest

141 718 Figure 1. 719 Downloaded from http://aem.asm.org/

720 on December 24, 2018 by guest

142 721 Figure 2. Downloaded from http://aem.asm.org/

722

723 on December 24, 2018 by guest

143 724 Figure 3. Downloaded from

725 http://aem.asm.org/

726 on December 24, 2018 by guest

144 727 Figure 4. Downloaded from http://aem.asm.org/ on December 24, 2018 by guest

728

729

145 730 Figure 5. Downloaded from http://aem.asm.org/ on December 24, 2018 by guest

731

732

146 733 Figure 6. Downloaded from http://aem.asm.org/ on December 24, 2018 by guest

734

147 735 Figure 7. Downloaded from http://aem.asm.org/ on December 24, 2018 by guest

736

148 Environ Sci Pollut Res DOI 10.1007/s11356-015-4947-z

RESEARCH ARTICLE

Characterization of the microbial community structure and the physicochemical properties of produced water and seawater from the Hibernia oil production platform

C. William Yeung 1,2 & Kenneth Lee3 & Susan Cobanli4 & Tom King 4 & Jay Bugden4 & Lyle G. Whyte2 & Charles W. Greer1

Received: 9 March 2015 /Accepted: 23 June 2015 # Springer-Verlag Berlin Heidelberg 2015

Abstract Hibernia is Canada’s largest offshore oil platform. in seawater up to 1000 m from the platform. These methods Produced water is the major waste byproduct discharged into could be used to track the dispersion of produced water into the ocean. In order to evaluate different potential disposal the surrounding ocean. methods, a comprehensive study was performed to determine the impact from the discharge. Microorganisms are typically Keywords Produced water . Petroleum waste byproduct . the first organisms to respond to changes in their environment. Microbial communities . DGGE . q-PCR . The objectives were to characterize the microbial communi- Thermoanaerobacter ties and the chemical composition in the produced water and to characterize changes in the seawater bacterial community around the platform. The results from chemical, physicochem- Introduction ical, and microbial analyses revealed that the discharge did not have a detectable effect on the surrounding seawater. The The Hibernia oil production platform is the largest offshore seawater bacterial community was relatively stable, spatially. oil production platform in Canada and is situated on top of Unique microorganisms like Thermoanaerobacter were one of the largest oilfields in Canada. The platform is cur- found in the produced water. Thermoanaerobacter-specific rently producing approximately 220,000 barrels of crude q-PCR and nested-PCR primers were designed, and both oil per day (http://www.hibernia.ca). Along with oil methods demonstrated that Thermoanaerobacter was present production, produced water is the largest volume of waste byproduct discharged into the ocean under Canadian ’ Responsible editor: Robert Duran regulations (Canada s Revised Offshore Waste Treatment Guideline 2002, http://www.cnsopb.ns.ca/sites/default/ Electronic supplementary material The online version of this article files/pdfs/owtg_redraft.pdf). Produced water consists of (doi:10.1007/s11356-015-4947-z) contains supplementary material, which is available to authorized users. formation water (water trapped with oil and gas in the reservoir) and seawater injected during the drilling * C. William Yeung process to maintain reservoir pressure. The chemical [email protected] compounds present in the produced water are basically the same as those present in the most water-soluble fraction 1 Energy, Mining and Environment, National Research Council of the crude oil in the reservoir, which includes natural Canada, 6100 Royalmount Ave., Montreal, QC H4P 2R2, Canada organics (aromatic and aliphatic hydrocarbons, organic 2 Department of Natural Resource Sciences, McGill University, acids, phenols), metals, and traces of chemicals added dur- Ste-Anne-de-Bellevue, QC H9X 3V9, Canada ing oil production (Røe Utvik et al. 1998). A large propor- 3 Oceans and Atmosphere National Research Flagship, Australian tion of these compounds is potentially toxic, genotoxic, and Resources Research Centre, CSIRO, 26 Dick Perry Avenue, carcinogenic to aquatic organisms (Jha 2004, 2008). Dis- Kensington, WA 6151, Australia charge of these chemicals could pose an environmental 4 Fisheries and Oceans Canada, PO Box 1006, Dartmouth, NS B2Y concern, and certain chemicals like the hydrocarbons pose 4A2, Canada an even greater environmental concern because of their

149 Environ Sci Pollut Res toxicity, potential for bioaccumulation, and persistence in offshore platforms. The results revealed that both bacterial the marine environment (Neff 1987, 2002). Paine et al. and archaeal community structures were relatively stable (1992) found that even after short-term exposure, hydrocar- within a 50-km region around the production platform, bons at concentrations as low as 1.3 mg/L had lethal effects suggesting that any changes in this environment could on fish larvae and sublethal effects at concentrations as low be an indication of the impact from produced water. as 0.13 mg/L. Although the discharge concentration is gen- Furthermore, a wide range of unique microorganisms have erally much lower (Neff 2002), the discharge volume is been identified by culture-dependent and culture-independent usually high and has a tendency to increase during the life methods from samples of produced water obtained from geo- of the well. This large volume of pollutants, even at low graphically distinct oil reservoirs throughout the world concentrations (in μg/L range), could still have long-term (Grassia et al. 1996), predominately in the North Sea (Dahle effects, and therefore requires monitoring at the contami- et al. 2008), California (Orphan et al. 2000, 2003), China (Li nant and biological response levels. Emerging evidence et al. 2006, 2007), Japan (Kobayashi et al. 2012), Siberia from the Baltic Sea and the North Sea investigations sug- (Bonch-Osmolovskaya et al. 2003; Davydova-Charakhch’yan gested that oil production activities may have an impact on et al. 1992), Brazil (Piubeli et al. 2014), and Western Canada fish and larvae at even greater distances from the platforms (Grabowski et al. 2005; Voordouw et al. 1996). The microor- (Stagg and McIntosh 1996; Rybakovas et al. 2009). There- ganisms identified from these studies has helped to not only fore, a thorough characterization of the Hibernia-produced improve our understanding of petroleum microbiology but water and the surrounding marine environment is necessary also to develop new environmental and industrial applica- to estimate the potential environmental impact. A number tions, such as oil spill remediation (Prince et al. 1999)and of studies (Brooks et al. 1980; Rabalais et al. 1991; Terrens microbial enhanced oil recovery (Banat et al. 2000; Banat and Tait 1996) found that the concentration of benzene de- 1995). creased from 50,000- to 150,000-fold just 20 m away from Despite all of the environmental and economic interests in the produced water discharge point, suggesting that using the offshore oil and gas reserves in Eastern Canada from both the current chemical detection methods, there would be government and industry, relatively limited knowledge is very little evidence of harmful chemical concentrations in available on the microbial diversity in the marine environment the marine environment from the produced water discharge. around the production platforms and subsurface ecosystems. Some studies that used fish (Abrahamson et al. 2008; Therefore, a more comprehensive characterization of the mi- Hylland et al. 2008;Brooksetal.2011) or shellfish crobial community structure in the produced water from this (Hylland et al. 2008;Brooksetal.2011) to monitor the petroleum-rich region is much needed and long overdue. The long-term effects of produced water in the surrounding en- molecular techniques, in particular, clone library and DGGE vironment have shown that exposure levels were generally analysis of the 16S rRNA genes, have been shown to be ef- low and caused minor biological impacts at the deployment fective in providing a more complete characterization of com- locations over the experimental time periods. In order to plex microbial assemblages in environmental samples monitor chronic effects, a better understanding of how the (Amann et al. 1995). chemicals and the other components in the produced water In this study, the first objective was to characterize the effluent are transported and diluted in the surrounding ma- unique bacterial and archaeal communities in the produced rine environment is required. It is also necessary to develop water and the bacterial communities in the surrounding sea- more sensitive methods to monitor the components in pro- water using 16S rRNA gene clone libraries, DGGE, and/or duced water at the concentrations that are found in the sur- sequencing analysis. Using the information from the commu- rounding ecosystem. nity characterization, the other objective was to develop more Marine bacterioplankton are central mediators of many sensitive methods to monitor the discharge of produced water, oceanic biogeochemical processes with abundances that in this case using a tracker organism derived from the pro- far exceed other living organisms (Azam and Malfatti duced water. The combination of these results might help de- 2007). Since microorganisms are typically the first organ- fine the extent of potential impacts from the discharge of pro- isms to encounter and respond to changes (i.e., chemical, duced water. thermal, etc.) in their environment at the population level, close monitoring of this ubiquitous biological component may provide a sensitive indicator of water quality. Yeung Materials and methods et al. (2011b) proposed using molecular biology methods, like denaturing gradient gel electrophoresis (DGGE), to Sample collection survey the 16S ribosomal RNA (rRNA) gene from the seawater microbial community as a monitoring method Seawater samples were collected in two different years. In to define the extent of produced water discharge around July 2005, seawater samples were collected using a Seabird

150 Environ Sci Pollut Res

Niskin rosette frame (24×10 L bottles) containing Seabird analysis, 2 l of the produced water were immediately fil- conductivity, temperature, and depth detectors (CTD) at tered through sterile 0.22 μm GSWP (Millipore) filters three different depths (2 m, 50 m, and near bottom (NB; stored in 50 mL tubes at −20 °C until analyzed. 0.5 m off the bottom)) from various locations (S0 (46.744, −48.7808), 500 m South; S1 (46.74133, −48.782), 1000 m South; S3 (46.72283, −48.7848), 3000 m South; Nutrients analyses S6 (46.572, −48.7817), 20 km South; N0 (46.7565, −48.7808), 500 m North; NW0 (46.75433, −48.7875), Silicates The determination of soluble silicates in seawater 500 m Northwest; W0 (46.7505, −48.7903), 500 m West) was based on Technicon Industrial Method 186-72W and a reference seawater was collected at a reference lo- (Strickland and Parsons 1973)byreductionof cation 50 km (R50K (46.6968, −49.4243)) west of the silicomolybdate in acidic solution to ‘molybdenum blue’ by Hibernia production platform at all three depths. In Ju- ascorbic acid. Details of the method are found in Lee et al. ly 2008, seawater samples were collected, using a Niskin (2011). bottle attached to a CTD which was deployed by hand from one of the ship’s launches, from a number of loca- Nitrate/nitrite The determination of nitrate/nitrite in seawater tions (50, 100, 200, 300, 500, 1000, and 2000 m) at two was based on Technicon Industrial Method 158-71W depths (1 and 50 m) from the south of the Hibernia plat- (Armstrong et al. 1967) by reducing nitrate to nitrite by a form and two reference seawater samples were collected copper-cadmium reductor column. Details of the method are at the same two depths (1 and 50 m) at a reference loca- found in Lee et al. (2011). tion 50 km (R50k) west of the Hibernia production platform (46.69699, −49.4266). All containers used in the Ortho-phosphate The determination of ortho-phosphate filtration were rinsed with the sample water three times. was based on Method 155-71W (Murphy and Riley In 2005, surrounding seawater was collected in two dif- 1962) to measure the formation of phosphomolybdenum ferent types of containers: acid rinsed 10 L Nalgene® blue complex produced by the reaction of phosphate with HDPE jerricans and solvent rinsed 4 L amber glass bottles an acidic ammonium molybdate solution containing a (for organic chemical analysis). Samples were stored at small amount of antimony and ascorbic acid. The original 4 °C on the research ship until processed. For organic method called for combining ammonium molybdate, anti- chemical analyses, samples stored in amber glass bottles mony potassium tartrate, and ascorbic acid into one work- were used. From the jerrican samples, an aliquot was re- ing reagent. In house, the ascorbic acid is introduced into moved for the measurement of nutrients (silicate, nitrate, the sample stream separately. nitrite, ammonia, and phosphate) and bacterial community analyses. In 2008, only jerrican samples were collected. Ammonia The method for determination of ammonia For microbial analysis, 4 L of seawater were immediately (Kerouel and Aminot 1997) was based on the reaction of filtered through sterile 0.22 μm GSWP (Millipore) filters. ammonia with ortho-phthaldialdehyde (OPA) and sulfite. De- Following filtration, all filters were transferred to sterile tails of the method are found in Lee et al. (2011). 50 mL Falcon tubes and stored at −20 °C until analyzed. In addition, samples of fresh produced water were Organic chemical analyses kindly provided by the personnel of the Hibernia production platform in 2005 and 2008. All produced water sam- Polycyclic aromatic hydrocarbons (PAHs) and aliphatic ples were collected at the well head at 1 h intervals into hydrocarbons This method was based on a modified version two different types of containers: acid-rinsed 10-L of EPA Method 8270. Details of the method are found in Nalgene® HDPE jerricans and solvent-rinsed 4-L amber Adams et al. (2014). glass bottles. All containers were rinsed three times with produced water, then filled completely without headspace, Alkylated and nonyl phenols Phenols were analyzed accord- sealed and transported to the research ship for further ing to a modified version of EPA method 8041. Method de- processing. After collection, samples were transferred tails are found in Lee et al. (2011). from the rig to the supply vessel. Once onboard the supply vessel, an aliquot was removed for the measurement Benzene, toluene, ethylbenzene, and xylene All benzene, of temperature, pH, salinity, nutrients (silicate, nitrate, ni- toluene, ethylbenzene, and xylene (BTEX) samples were an- trite, ammonia, and phosphate), and bacterial community alyzed within 2 weeks of collection. For the analysis of analyses. Four (4) L samples were stored at 4 °C for BTEX, EPA Method 8240 (purge and trap) was modified by chemical analysis and a Nalgene jerrican sample was used running the GC/MS in selected ion monitoring mode. Details immediately for microbial analysis. For microbial of the method protocols are found in Conmy et al. (2014).

151 Environ Sci Pollut Res

Genomic DNA extraction EstimateS;Colwell2005). The Shannon evenness index was calculated using the formula E=eH′/N,whereH′ is the Total community DNA from both seawater and produced Shannon index of diversity and N is the total number of water were extracted from the filters using the method phylotypes (Krebs 1989). described by Fortin et al. (1998) with minor modifications. Details of the method protocols are found in Yeung et al. Molecular analysis for produced water and seawater (2011b). samples Total community DNA from the 4-L seawater, 4-L produced water, and 1 L of each seawater/produced water dilution PCR amplification of the bacterial 16S rRNA genes, DGGE series from 2008 were extracted from the filter using an Ul- analysis, dendrogram analysis, and phylogenetic analysis traClean® Water DNA isolation kit (Mo Bio Laboratories, were preformed as previously described (Yeung et al. 2010, Carlsbad, CA) following the manufacturer’s protocol, except 2011a). Archaea 16S rRNA gene were PCR amplified using the DNA was eluted into 3 mL of elution buffer at the final the archaea-specific forward primer ARC344F (5′-ACGG step. GGYGCAGCAGGCGCGA-3′) with the same GC clamp se- The DNA concentration from all extractions was estimated quence at the 5′ end as for the bacteria-specific forward primer by agarose gel electrophoresis using 5 μL of purified material and the reverse primer ARC915R (5′-GTGCTCCCCCGGCA against the Lambda HindIII DNA ladder (Amersham Biosci- ATTCCT-3′) which generates a 572-bp fragment. In brief, ences, Piscataway, NJ) standard on a 0.7 % agarose gel stained each 50 μL PCR mixture contained ~1 ng/μL of the template with SYBR safe. DNA, 25 pmol of each oligonucleotide primer, 200 μMof

each dNTP, 1 mM MgCl2, 2.5 units of Taq polymerase Clone libraries analysis for 2005 produced water (Amersham Biosciences, Piscataway, NJ, USA), and 1× Taq polymerase buffer (10 mM Tris–HCl (pH 9.)0, 50 mM KCl,

The extracted DNA from the produced water was used to 1.5 mM MgCl2). Briefly, after an initial temperature of 96 °C construct two 16S rRNA gene clone libraries, one bacterial for 5 min and thermocycling at 94 °C for 1 min, the annealing and one archaeal. To construct each library, three PCR repli- temperature was set to 60 °C for 1 min and decreased by 1 °C cates per sample were combined to minimize bias. Bacterial every cycle for 10 cycles, with a 3-min elongation time at and archaeal 16S rRNA genes were PCR amplified using two 72 °C. Additional cycles (15–20) were performed with an- sets of primers: F1 (5′-GAGTTTGATCCTGGCTCAG-3′) nealing temperatures at 50 °C. The presence, size, and quan- and R13 (5′-AGAAAGGAGGTGATCCAGCC-3′)(Liesack tity of the PCR products were determined using agarose gel as et al. 1991) and A2F (5′-TTCCGGTTGATCCYGCCGGA- previously described (Yeung et al. 2010, 2011a). 3′) (Reysenbach and Pace 1995; Martinez-Murcia et al. The 16S rRNA gene products from four to eight PCR re- 1995) and A1406R (5′-GACGGGCGGTGTGTRCA-3′) actions were combined for each sample and concentrated by (Hansen et al. 1998; Reysenbach and Pace 1995), respective- ethanol precipitation before DGGE analysis. About 550 ng of ly. PCR amplification, cloning, and transformation were the 16S rRNA gene product from each sample was applied to performed as described by Yeung et al. (2011c) with the fol- a lane and analyzed on 8 % polyacrylamide DGGE gels con- lowing minor modification. First, archaeal cycle conditions taining gradients of 30–70%denaturant(asolutionwith7M were an initial denaturation temperature of 96 °C for 5 min, urea and 40 % deionized formamide was considered to be then 30 cycles at 94 °C for 1 min, 56 °C for 1 min, and 72 °C 100 % denaturant). DGGE, dendrogram, and phylogenetic for 2 min. Second, Escherichia coli strain SURE cell analyses were performed as previously described (Yeung (Invitrogen) was used for the transformation. et al. 2010, 2011a). A total of 106 white colonies from the bacterial library and 95 white colonies from the archaeal library were randomly q-PCR and nested-PCR primer design picked, transferred onto ampicillin-supplemented LB plates, and then incubated overnight. Transformant selection, se- All primers used in this study were designed using PRIMER quencing, phylogenetic analyses, and rarefraction calculation 3.0 (http://frodo.wi.mit.edu/primer3/; Rozen and Skaletsky were preformed as described in Yeung et al. (2011c). The 2000) based on the 16S rRNA gene sequences of the Shannonindex(H′) of diversity (Magurran 2004), the Thermoanaerobacter clones from the produced water clone Simpson’s reciprocal index of dominance (Simpson’sindex library (JF89561) and a number of Thermoanaerobacter se- of dominance (D) is the inverse of the Simpson’sreciprocal quences from GenBank matches (AY701757, AY701758, index of dominance) (Magurran 1988, 2004; Hayek and AY701759, EF026571, Y11279). All sequences were first Buzas 1996; Simpson 1949), and the bias-corrected Chao1 aligned using the MacVector 9.0 software package (Accelrys, estimator of total species richness (Chao 1984) were deter- Cary, NC, USA) to create a consensus sequence. Primer de- mined using EstimateS 8.2 (http://viceroy.eeb.uconn.edu/ sign was carried out with the PRIMER 3.0 program using the

152 Environ Sci Pollut Res consensus sequence. A pair of primers (TMF1: CCGTAGCG each run to assess reaction-to-reaction variability. Three serial AACGCAATAAGT and TMR1: CTGTGCAGGCTCCTTA dilutions of each test sample were used for q-PCR amplifica- CCTC) targeting a 201-bp fragment of the 16S rRNA gene tion to determine the best concentration range. Reaction effi- from the consensus sequence was designed and used for both ciency ([10(−1/M)]−1) and amplification value (10(−1/M)), q-PCR and the second part of the nested-PCR. Another pair of where M=the slope of the standard curve, were calculated as primers (TMfull-F: TGTAGCGGTGAAATGCGTAG and per the Rotor-Gene manual. An ideal amplification value TMfull-R: ACCTTCCGATACGGCTACCT) targeting a lon- should be close to 2, and the ideal reaction efficiency should ger fragment (849 bp) of the 16S rRNA gene from the con- be close to 1. The q-PCR assay also produced a threshold sensus sequence was designed for the first part of the nested- cycle (CT) value, which is the number of cycles at the point PCR. The specificity of the primers was tested by submitting where fluorescence rises prominently above background. the primer sequences to the PROBE MATCH (Ribosomal Database Project II; Maidak et al. 2001) and BLAST pro- Nested-PCR amplification grams (http://blast.ncbi.nlm.nih.gov/Blast.cgi; Altschul et al. 1990) before laboratory testing. Similar to the optimization of q-PCR primers, the annealing temperature for the nested-PCR primer sets was calculated q-PCR amplification using their ATGC content and tested with a temperature gradient PCR, where 57 °C was found to be the optimum tem- For laboratory testing and calibration curves for testing the q- perature. Primer specificities were first assessed using the reg- PCR primers, plasmid DNA containing cloned target se- ular PCR against the 16S rRNA gene sequences of the quences was used. This method had been used in a number Thermoanaerobacter clones from the produced water clone of studies to create a standard curve (Galluzzi et al. 2004; library (JF89561) and against Bclean^ surrounding seawater Wawrik et al. 2002; Zhu et al. 2005). It is also the only avail- DNA to check for non-specific binding. The first part of PCR able method for calibrating q-PCR detection of uncultured was performed with a 50 μL PCR mixture containing 2.5– taxa (Suzuki et al. 2000). The advantage of calibrating with 5 ng of template DNA, 25 pmol of each oligonucleotide prim- plasmid DNA is that the exact number of target genes can be er (TMfull-F and TMfull-R), 200 μMofeachdNTP,1mM calculated, knowing the concentration of the plasmid DNA MgCl2, and 2.5 units of Taq polymerase (Amersham Biosci- standard. A plasmid DNA containing the 201-bp sequence ences, Piscataway, NJ, USA) in Taq polymerase buffer was manufactured from MiniGenes Custom Gene Synthesis (10 mM Tris–HCl (pH 9.0), 50 mM KCl, 1.5 mM MgCl2). (IDT, Coralville, IA). Annealing temperature for the primer set The PCR was performed with a predenaturation at 96 °C for was first calculated with their ATGC content and tested with a 5 min followed by 25 cycles of denaturation at 94 °C for temperature gradient PCR and q-PCR, where 61 °C was found 1 min, annealing at 57 °C for 1 min, and extension at 72 °C to be the optimum temperature. Primer specificities were first for 2 min. The second part of the PCR was performed with the assessed using regular PCR and then q-PCR against the plas- same PCR mixture but with 5 μL of the product from the first mid DNA, and finally they were tested against genomic DNA PCR reaction as template and the q-PCR forward and reverse from Bclean^ surrounding seawater with and without plasmid primers (TMF1 and TMR1). The PCR conditions were similar DNA to check for non-specific binding. to q-PCR amplification. Briefly, initial denaturation was at The q-PCR reaction mixture followed the manufacture’s 95 °C for 15 min, followed by 35 cycles of 95 °C for 10 s, instructions. In brief, in each sample, a 20-μL reaction mixture 61 °C for 15 s, and 72 °C for 15 s. consisted of 10 μL of QuantiTect SYBR Green PCR Master Mix (including HotStarTaq DNA polymerase, 5 mM MgCl2, Nucleotide sequence accession numbers dNTP mix, QuantiTect SYBR Green PCR Buffer), 0.4 μLof each 25 μM of each of TMF1 and TMR1 primers, 0.8 μLof The 16S rRNA gene sequences obtained in this study were

25 mM MgCl2,3.4μL of RNase-free water, and 5 μLtem- deposited in the GenBank database under accession numbers plate DNA (~0.5 ng/μL). Real-time PCR assays were con- JF789483 to JF789518 and JF789541 to JF789563. ducted using a Rotor-Gene 3000 Real-Timer PCR machine and monitored with the Rotor-Gene 6 software (Corbett Re- search, San Francisco, CA). The cycling program consisted of Results an initial denaturation at 95 °C for 15 min, followed by 50 cycles of 95 °C for 10 s, 61 °C for 15 s, and 72 °C for 15 s. The Physicochemical and chemical analyses of the produced q-PCR results were analyzed with Rotor-Gene 6 software. water and seawater All q-PCR experiments were performed with triplicate 10−6 sample dilutions from the standard curve as positive con- The Hibernia-produced water temperature was around 82 °C, trols and a triplicate of negative controls (sterile water) for with the pH at 7.78, and the salinity was 45.6 parts per

153 Environ Sci Pollut Res thousand (ppt). In contrast, the temperature and salinity were Epsilon-proteobacteria were Arcobacter, which comprised much lower in the surrounding seawater (Table 1). Other the entire group with 21.7 % of the clone library. However, physicochemical characteristics for the produced water and the most dominant phylotype in the bacterial library was seawater are listed in Table 1. The results showed that several Thermoanaerobacter, from the Firmicutes. Firmicutes was chemicals (i.e., silicate, ammonia, and various petroleum hy- the second largest phylum in the library with 34.9 % of the drocarbons) in the Hibernia-produced water occur at concen- clones. Within this phylum, Thermoanaerobacter by itself trations at least hundreds of times greater than those in the comprised 25.5 % of all the clones in the library. The other surrounding seawater (Table 1) and thus could potentially be phylum was Deferribacteres, represented by Flexistipes. used as natural tracers to track the discharge of produced wa- Overall, 10 out of 23 phylotypes (68 clones) were related to ter. In particular, the produced water had a variety of petro- anaerobic bacteria or facultative anaerobes, which consisted leum hydrocarbons: 4100±128 μg/L BTEX, 11.9±1.1 μg/L of 64.2 % of the total number of clones. Within the anaerobe phenols, 33.4±0.8 μg/L alkanes, and 116.4±17.9 μg/L of genera, nine (45 clones) were related to thermophilic genera, PAHs (Table 1). However, the seawater chemical analysis which consisted of 43 % of the clones. The remaining clones revealed that the PAHs, phenols, alkanes, silicate, phosphate, consisted of a diverse number of phylotypes (13 out of 23) and ammonia were all either not detected or at very low con- with fewer clones (35.8 %), and they were all related to centrations, similar to the reference seawater (Table 1), sug- mesophilic aerobic Alpha- and Gamma-proteobacteria. gesting that although they were discharged at high volume and high concentration, all these chemicals were diluted to below detection levels within 500 m of the production platform. Produced water archaeal clone library analysis

The archaeal clone library had a lower diversity than the bac- Produced water bacterial clone library analysis terial clone library. The archaeal clone library comprised 95 clones that were grouped into only three phylotypes (Fig. 2). The produced water bacterial clone library was composed of All of the phylotypes showed at least a 99 % match to the 106 clones that grouped into 23 phylotypes (Fig. 1). Most of available known cultured sequences from GenBank. All the the phylotypes showed at least a 97 % match in GenBank to a phylotypes belonged to the Euryarchaeota. The phylotypes known cultured bacterium. The phylotypes could be divided could be further divided into two classes: Thermococci and into three phyla as follows: Proteobacteria (60.4 % of the total Archaeoglobi.TheThermococci was the most dominant phy- clones), Firmicutes (34.9 %), and Deferribacteres (4.7 %). lum (96.8 % of the total clones) and was represented by two The majority of the phylotypes (16 out of 23) belonged to different phylotypes closely related to species of the Proteobacteria, which had representatives from the Gam- Thermococcus: Thermococcus litoralis and Thermococcus ma-proteobacteria (25.5 % of the total clones), Epsilon- alcaliphilus. Only a very small percentage of the clones proteobacteria (21.7 %), Alpha-proteobacteria (10.4 %), (3.2 %) were very closely related to Archaeoglobus fulgidus and Delta-proteobacteria (2.8 %). The most dominant from the Archaeoglobi.

Table 1 Physicochemical and chemical characteristics of Produced water Average surrounding seawater Hibernia-produced water and seawater; the first number is the Temperature ~82 °C ~2.62 °C average and the second number is pH 7.78 N.A. the standard deviation Salinity 45.6 ppt ~32.7 ppt Silicate 852.04±89 μM2.5μM Nitrate 0.34±0.005 μM1.95μM Nitrite 0.34±0.05 μM0.36μM Ammonia 641.49±0.38 μM5.26μM Phosphate 12.08±3.62 μM0.72μM Total sulfur 726±2 mg/L N.D. Total BETX 4100±128 μg/L N.D. Total dissolved PAHs 116.4±17.9 μg/L N.D. Total alkanes 33.4±0.8 μg/L N.D. Total phenols 11.9±1.1 μg/L N.D.

N.D. not detected, N.A. not analyzed

154 Environ Sci Pollut Res

Fig. 1 Phylogenetic relationship of the 23 bacterial 16S rRNA gene of sequence from each clone. Aquifex pyrophilus wasusedasthe sequences obtained from Hibernia-produced water clone library outgroup. Numbers on the nodes are the bootstrap values based on (HibPWCl). The clones were labeled with HibPWCl- and with bacteria 1000 replicates. The scale bar indicates the estimated number of base (BAC) and a number. The tree was inferred by neighbor-joining analysis changes per nucleotide sequence position

155 Environ Sci Pollut Res

Fig. 2 Phylogenetic relationship of the three archaeal 16S rRNA gene of sequence from each clone. Aquifex pyrophilus wasusedasthe sequences obtained from Hibernia-produced water clone library outgroup. Numbers on the nodes are the bootstrap values based on (HibPWCl). The clones were labeled with HibPWCl- and with archaea 1000 replicates. The scale bar indicates the estimated number of base (ARC) and a number. The tree was inferred by neighbor-joining analysis changes per nucleotide sequence position

Statistical analysis of clone libraries Comparatively, the diversity of the bacterial clone library was much higher (23 phylotypes) than the diversity The number of clones, phylotypes, and biodiversity indi- of the archaeal library (3 phylotypes). This is supported ces calculated for the two clone libraries are summarized by the higher value of the Shannon index (2.45 for bac- in Table 2. The coverage of both clone libraries was high terial, 0.51 for archaeal) and the higher value of the ranging from 91.5 to ~100 %, suggesting that the majority Simpson’s reciprocal index (7.77 for bacterial, 1.38 for of the microbial diversity was identified in this study. The archaeal) (Table 2). Even with the higher diversity, the high bacterial clone library coverage value corresponded bacterial population distribution is more even than the to the near-plateau rarefaction curve (data not shown) and archaeal. The bacterial clone library Shannon evenness the Chao1 diversity estimation result was 29 (Table 2). index was 0.78 (with 1 being even) compared with the The Chao1 value estimated that a total of 29 phylotypes archaeal clone library index value of 0.46. The uneven- could potentially be expected from this environment, of ness of the archaeal population distribution suggested a which 23 were identified. For the archaeal clone library, high species dominance (the Thermococci dominated at the very high coverage value (~100 %) was supported by 96.8 % of the clones), which is also supported by the the rarefaction curves that almost reached a plateau (data higher value of the inverse Simpson’s index of dominance not shown). Furthermore, the total number of phylotypes (0.72: with 1 being no diversityand0beinginfinitedi- (3) is equal to the estimation from the Chao1 diversity versity) (Table 2). analysis (Table 2). Produced water bacterial and archaeal DGGE analysis

Table 2 Bacterial and archaeal clone library analysis The DGGE analysis for the bacterial and archaeal 16S rRNA genes was intended to identify the dominant microbial groups Bacterial Archaeal in the composition of the produced water, so all the major Number of clones 106 95 bands were excised and re-amplified for sequencing analysis. Number of phylotypes 23 3 The bacterial DGGE gel displayed a much higher number of Percentage coverage 91.5 % ~100 % bands than the archaeal DGGE gel (only two major bands) Shannon index 2.45 0.51 (Fig. 3). From the bacterial DGGE, a total of seven DGGE Shannon Evenness index 0.78 0.46 bands were excised and sequenced (Fig. 3). All of the se- Simpson index 7.77 1.38 quences showed at least a 97 % match to the available se- Inverse Simpson’s index (D) 0.13 0.72 quences of known cultured bacteria from GenBank. Even with Chao1 29 3 just seven sequences, they all fall into a diverse number of groups: the most dominant are from Alpha-proteobacteria

156 Environ Sci Pollut Res

Seawater bacterial DGGE analysis

Seawater bacterial community structure analysis revealed that the community from the surrounding seawater clustered closely by depth with two major clusters: one cluster for samples from the surface to 25 m and the other cluster for samples from

50mtoNB(SAB ≈67) (Fig. 6). Much higher SAB values (higher similarity) were observed with distance from the production platform, which suggested that the variation in the bacterial community structure is more likely related to depth than by distance from the point of produced water discharge (Figs. 6 and 7). Similar to the chemical analyses (Table 2), the microbial analysis revealed that the bacterial community structure from the produced water is very different from that

of the surrounding seawater (SAB=38.8, Fig. 6), even though the major component of the produced water comes from the injected surrounding seawater, suggesting that the bacterial community structure in the produced water is unique. The seawater DGGE analysis also revealed that there is a higher bacterial diversity in the seawater (Fig. 7) than the produced water (Fig. 3). A total of 24 DGGE bands were excised and sequenced from the seawater DGGE analysis. Almost all the sequences were at least a 97 % match to the existing sequences from GenBank and showed high similarities to GenBank sequences from bacteria originating from either the Arctic Sea or other Atlantic regions (Fig. 8). The sequences fall equally into only two groups: Proteobacteria and Bacteroidetes (with 12 sequences each). Within the Proteobacteria, the 12 sequences were further divided into two subclasses: Alpha-proteobacteria (seven sequences) and Gamma-proteobacteria (five sequences). Bands Hib05-SW4, 6, 7, 8, 9, 11, 12, 16, 17, and 22 were present only in the bottom water samples (50 m and NB) (Fig. 7), suggesting that some species were unique to the lower water column. These sequence differences were more closely related to depth than to distance from the platform, suggesting that the seawater depth had a more significant effect on the Fig. 3 Hibernia-produced water 16S rRNA gene DGGE fingerprint. a community structure than the distance from the produced wa- Bacterial DGGE fingerprint, bands were labeled from 1 to 7. b Archaeal ter discharge. DGGE fingerprint, bands were labeled from 1 to 2 q-PCR analysis and Gamma-proteobacteria (both were represented by two out of seven sequences), and one sequence from each of the Since the Thermoanaerobacter sp. from the Hibernia- Epsilon-proteobacteria, Firmicutes,andThermococci produced water is uncultured, plasmid DNA was the only (archaeal: from bacterial primer mismatches (data not method that could be used to produce the standard curve. By shown)) (Fig. 4). optimizing q-PCR conditions, we obtained a standard curve Similar to the clone library results, the archaeal DGGE with a linear range across six 10-fold dilutions of DNA con- showed lower diversity than the bacterial DGGE. Only two centrations (Fig. S1a). The standard curve was plotted with the intense bands were excised and sequenced (Fig. 3). One of the CT against the copy number of Thermoanaerobacter plasmid sequences was closely related to Thermococcus sp., similar to DNA (Fig. S1b). The detection limit was 21 copies in our the one that was identified from the bacterial DGGE, and the standard curve (Fig. S1). The standard curve indicated a good other band produced a sequence closely related to correlation between the amount of template (number of cop- 2 Archaeoglobus sp. (Fig. 5). ies) and the amount of product (represented by CT)(r =

157 Environ Sci Pollut Res

Fig. 4 Phylogenetic relationship of the seven bacterial 16S rRNA gene of sequence from each clone. Aquifex pyrophilus wasusedasthe sequences obtained from Hibernia-produced water (HibPW). The bands outgroup. Numbers on the nodes are the bootstrap values based on were labeled with HibPW- and with bacteria (BAC) and their band 1000 replicates. The scale bar indicates the estimated number of base numbers from Fig. 3a. The tree was inferred by neighbor-joining analysis changes per nucleotide sequence position

0.99901). The linearity of the standard curve and the fact that using just the plasmid DNA to determine the number of copies the PCR operates with constant efficiency confirmed that the of the gene, serial dilutions with various ratios of produced assay was well suited to quantitative measurements of water in natural seawater were used to determine the extrac- Thermoanaerobacter from environmental samples. tion efficiency from high to low produced water concentra- However, one of the disadvantages of using plasmid DNA tions and to determine the correlation between dilution and standards are the uncertainty of the DNA extraction efficiency number of copies of the gene. A total of 318,339.7 copies of from environmental samples and the potential presence of Thermoanaerobacter were present in the 5 μLofextracted multiple copies of the 16S rRNA gene. Therefore, instead of genomic DNA from 4 L of raw produced water (i.e., ~47,

Fig. 5 Phylogenetic relationship of the two archaeal 16S rRNA gene analysis of sequence from each clone. Aquifex pyrophilus was used as sequences obtained from Hibernia-produced water (HibPW). The bands the outgroup. Numbers on the nodes are the bootstrap values based on were labeled with HibPW- and with archaea (ARC)andtheirband 1000 replicates. The scale bar indicates the estimated number of base numbers from Fig. 3b. The tree was inferred by neighbor-joining changes per nucleotide sequence position

158 Environ Sci Pollut Res

Fig. 6 2005 Hibernia seawater bacterial 16S rRNA gene DGGE fingerprint. Bands were labeled from 1 to 24

750,955 copies in 1 L of produced water). With this high The results (Table 3)revealedthatThermoanaerobacter was starting concentration of Thermoanaerobacter copies in the detected in almost all the samples from within 500 m, showing raw produced water, three 10-fold dilutions of the produced that components of produced water were detectable in the water with seawater were made (i.e., from 1/10 to 1/10,000). surrounding seawater within 500 m of the discharge. The We obtained a standard curve with a linear range (r2=0.9998) highest concentration of Thermoanaerobacter with ~26,740 across seven DNA concentrations (Fig. S2). The detection copies/L (178.27 copies in 5 μLofgenomicDNAfrom4Lof limit was about eight copies in 5 μL of the genomic DNA seawater sample) was found in seawater sampled at 100 m per q-PCR reaction (i.e., about 4775 copies/L), equal to a from the production platform at 50 m depth. Much lower 1:10,000 dilution (Fig. S2). The standard curves indicated a number of copies of Thermoanaerobacter was found in the good correlation between the amount of template (number of rest of the seawater samples (Table 3), and only an average of copies) and the dilution (r2=0.9998) (Fig. S2). The linearity of around 200 copies was found in samples collected from 1 m the standard curves and the fact that the PCR operates with depth (Table 3). constant efficiency between replicates, confirmed that the assay was well suited for quantitative measurements of the di- 2008 Hibernia seawater nested-PCR analysis lution of produced water in environmental samples. The nested-PCR result confirmed all the positive amplifica- 2008 Hibernia seawater DGGE analysis tions from the q-PCR analysis, and showed weak amplification from the 1000 m samples collected at 1 m depth (Table 3). Before applying the q-PCR method to the seawater samples, No amplification of Thermoanaerobacter was detected be- the bacterial community in the 2008 Hibernia surrounding yond 1000 m by nested-PCR (Table 3). seawater was analyzed the same way as the 2005 seawater samples. The DGGE results (Fig. S3a, b) also revealed that similar banding patterns were observed in the seawater sam- Discussion ples at the same depths (SAB≥95.5), suggesting that the bacterial community structure in the surrounding seawater was Physical and chemical characterization stable over time. of the Hibernia-produced water

2008 Hibernia seawater q-PCR analysis Since no two produced waters are the same, for a better understanding of transport and dilution, a detailed analysis of the The q-PCR method was used to estimate the number of copies physical and chemical characteristics of the Hibernia- of Thermoanaerobacter in the Hibernia surrounding seawater. produced water was required. Similar to other produced

159 Environ Sci Pollut Res

Fig. 7 Phylogenetic relationship of the 24 bacterial 16S rRNA gene Aquifex pyrophilus wasusedastheoutgroup.Numbers on the nodes are sequences obtained from Hibernia seawater DGGE. The bands were the bootstrap values based on 1000 replicates. The scale bar indicates the labeled with Hib05-SW and a number from Fig. 6.Thetreewas estimated number of base changes per nucleotide sequence position inferred by neighbor-joining analysis of sequence from each clone. waters from the Northern Atlantic region (like the North Sea discharge consists of dissolved and particulate inorganic and detailed in Tibbets et al. 1992; Flynn et al. 1995), the organic chemicals. The salinity of produced waters in general Hibernia-produced water is discharged at a relatively high ranges from a few parts per thousand to a saturated brine temperature (Table 1). This high-temperature, complex (~300 ppt) and is usually denser than seawater (Rittenhouse

160 Environ Sci Pollut Res

Fig. 8 Bacterial 16S rRNA gene DGGE fingerprint cluster analysis from Hibernia seawater and Hibernia-produced water

et al. 1969;Large1990; Collins 1975). The salinity of those found normally in other produced waters from around Hibernia-produced water is 45.6 ppt. Hibernia-produced wa- the world (Neff 2002). ter also had noticeably elevated concentrations of silicate, am- Produced water contains residual petroleum hydrocarbons, monia, phosphate, and various petroleum hydrocarbons which are the organic components of greatest environmental (Table 1). The Hibernia-produced water has higher ammonia concern. Produced water is generally pre-treated to remove and phosphate concentrations (Table 1) than those observed in dispersed oil before being discharged to the ocean. After treat- other produced waters but lower nitrate concentrations than ment, the remaining petroleum hydrocarbons are mainly low

Table 3 Surrounding seawater q-PCR and nested-PCR results

Sampling depth (m) Distance from platform

50 m 100 m 200 m 300 m 500 m 1000 m 2000 m R50K

1 310.5 258 256.5 208.5 139.5 N.D. N.D. N.D. q-PCR 50 11,995.5 26,740.5 3491.5 936 N.D. N.D. N.D. N.D. 1 + + ++++ −−Nested-PCR 50 + + + + −− − −

Thermoanaerobacter 16S rRNA gene copies per liter of seawater N.D. not detectable, B+^ detected, B−^ not detected

161 Environ Sci Pollut Res molecular weight aromatic hydrocarbons (like benzene, tolu- dilution factor chemical markers could not be used to define ene, ethylbenzene, and xylenes (BTEX)) and small amounts the area of impacts. The use of biological factors may provide of dissolved saturated hydrocarbons (alkanes), in addition to a better resolution. PAHs and phenols. BTEX are often the most abundant hydrocarbons in produced water, ranging from 0.068 to 578 mg/L in produced water world-wide (Neff 2002). Similarly, the highest Produced water bacterial community structure concentrations of hydrocarbons in Hibernia-produced water are BTEX at 4100 μg/L (Table 1). The second most abundant In this study, molecular analyses were used to characterize the hydrocarbons in the Hibernia-produced water are PAHs at bacterial diversity in the produced water and surrounding sea- 137 μg/L (Table 1) and are still in the concentration ranges water. First, although much effort has been made to define the that are found world-wide (Neff 2002). The Hibernia- microbial diversity of petroleum reservoirs (e.g., Grassia et al. produced water contains relatively low concentrations of al- 1996), this study was one of the few to do so in Canadian kanes and phenols (~40 and 12 μg/L, respectively) when com- offshore oil production areas. The results showed that pared with the concentrations that have been measured world- Hibernia-produced water harbors a diverse microbial commu- wide (Neff 2002). nity. All phylotypes were closely matched to microbes associated with oil reservoirs, produced water, or other similar Comparison of the produced water and surrounding marine environments around the world (Figs. 1 and 2). seawater The Proteobacteria were the most dominant phylum with over 60 % of the total clones and with representatives from The Hibernia-produced water, although consisting mainly of four out of five classes: Alpha-, Gamma-, Delta-,andEpsilon- injected seawater, is physically, chemically, and biologically proteobacteria. Almost all the phylotypes identified within distinct from the surrounding seawater (Table 1;Fig.6). Our the Alpha- and Gamma-proteobacteria were closely related results showed that several chemicals (e.g., silicate, ammonia, to previously identified mesophilic aerobic marine genera: and various petroleum hydrocarbons) in the Hibernia- Sphingomonas, Sulfitobacter, Loktanella, Paracoccus, produced water occur at concentrations at least hundreds of Acidovorax, Pseudoalteromonas, Glaciecola, Alteromonas, times greater than those in the surrounding seawater and thus Alcanivorax, Pseudomonas,andMarinobacter (Fig. 1). Since can potentially be used as natural tracers to track the discharge the Hibernia oil reservoir is continuously injected with seawa- of produced water. The most abundant hydrocarbons in ter, large numbers of mesophilic aerobic marine microorgan- Hibernia-produced water is BTEX, which is extremely vola- isms could be introduced into the reservoir. Some of these tile and is usually lost rapidly during the discharge (Terrens introduced microorganisms may reside in the cooler portions and Tait 1996). Therefore, it is understandable that these vol- of the oil production system (like the production piping), atile compounds would rapidly fall below detection limits in which may explain the detection of mesophilic bacteria in seawater samples outside 500 m from the production platform. the produced water (Orphan et al. 2000). Some studies have The second most abundant hydrocarbons in the Hibernia- shown that mesophilic microorganisms constituted a large produced water were the dissolved PAHs (Table 1), which fraction of microorganisms detected in produced water and pose the greatest environmental concern in produced water similar environments (Li et al. 2006; Magot et al. 1992; because of their toxicity, bioaccumulation and persistence Orphan et al. 2000, 2003). It is also important to note that (less volatile and degradable) in the marine environment (Neff within these mesophilic aerobic marine genera, most of them, 1987, 2002). Therefore, PAH concentration is a good param- like the Nazina et al. (1995) findings, were related to genera of eter for monitoring produced water discharge. As an example, known hydrocarbon degraders: Sphingomonas (Romine et al. Harman et al. (2009) assessed the environmental impact of the 1999), Acidovorax (Meyer et al. 1999), Pseudoalteromonas discharge of produced water by monitoring the dissolved (Hedlund and Staley 2006), Alteromonas (Iwabuchi et al. PAHs and alkylphenols (AP) noting that the concentrations 2002), Alcanivorax (Yakimov et al. 1998), Pseudomonas of these compounds were several orders of magnitude lower (Le Petit et al. 1975), and Marinobacter (Gauthier et al. than those reported to give both acute and sublethal effects. 1992). The sequences not related to any known hydrocarbon Similarly, our results found that PAHs, phenols and alkanes degrader isolates were still closely related to sequences from were not detected in the surrounding seawater outside of hydrocarbon-degrading microbial communities (e.g., 500 m (Table 1), suggesting that although they were Sulfitobacter (Brakstad and Lødeng 2005), Paracoccus discharged at concentrations higher than in the surrounding (Guo et al. 2005), and Glaciecola (Brakstad et al. 2008)). seawater, PAHs were diluted to below detection levels within These mesophilic bacteria might have survived in hot forma- 500 m of the production platform. These results suggested that tions, and the surviving mesophilic bacteria might have although the concentrations of various chemical components thrived in the cooler portion of the oil production tubing en- were high within the produced water, with the even higher vironment, using the various types of hydrocarbons in the

162 Environ Sci Pollut Res produced water (Table 2) from the crude oil as energy and/or Produced water archaeal community structure carbon sources. Two phylotypes closely related to Delta-proteobacteria Only three phylotypes were identified in the archaeal clone were identified but with less than three clones, which sug- library (Fig. 3). The archaeal sequences showed high similar- gests that they might not be the major components of the ities to the genera Archaeoglobus (A. fulgidus)and bacterial community. Also, as mentioned previously, the Thermococcus spp.: T. litoralis and T. alcaliphilus. A. fulgidus major component of produced water came from the is a hyperthermophilic sulphidogenic bacterium that was iso- injected surrounding seawater, so it is understandable that lated from a marine hydrothermal system (Stetter 1988)and Beta-proteobacteria were not found in the produced water, from a North Sea oil reservoir (Beeder et al. 1994). T. litoralis since they appear to be relatively rare in marine bacterial is a hyperthermophilic sulfate-reducing bacterium first isolat- communities (Hahn 2006). ed from a submarine thermal spring (Neuner et al. 1990). Most interestingly, Epsilon-proteobacteria were found T. alcaliphilus is also a hyperthermophilic sulfur-respiring as the second most dominant class within the bacterium isolated from a submarine thermal system (Keller Proteobacteria with almost a quarter of the total clones et al. 1995). Numerous Thermococcus species have been de- (21.7 %) (Fig. 1), and it is represented by a single phylo- scribed from high-temperature oil reservoirs around the world, type closely related to Arcobacter sp. Similar Arcobacter suggesting the indigenous origin of hyperthermophilic ar- spp. have been found in microbial communities in a high- chaea in the deep subsurface biosphere (Neuner et al. 1990; temperature offshore petroleum reservoir in the North Sea L’Haridon et al. 2002; Grassia et al. 1996; Orphan et al. 2000; (Kaster et al. 2009). Kaster et al. (2009) also found that Ren et al. 2011; Kobayashi et al. 2012). clones related to Arcobacter (sulfur compound-oxidizing Epsilon-proteobacteria) were enriched in acetogen and Comparison of physicochemical parameters fermentative cultures at 55 °C as well as in fermentative with the bacterial and archaeal community structure and sulfate-reducing media at 70 °C. Recently, Arcobacter spp. were also identified in a petroleum degrading wet- The Hibernia-produced water is hot and salty but has a neutral land soil from the Shengli Oil Field on the Yellow River pH (Table 1). In most thermal environments, oxygen is usual- Delta (Han et al. 2009) and in the production water in a ly limiting. The results revealed that 64.2 % of the total num- crude oil gathering and transferring system (Liu et al. ber of clones from the bacterial clone library and 100 % of the 2009) at much lower temperatures. Liu et al. (2009)re- total number of clones from the archaeal clone library were vealed that Arcobacter was only found in the producted closely related to known anaerobes (Figs. 1 and 2). Within water and not in the crude oil. This result suggests that these anaerobic genera, almost all of them (as shown in Figs. 1 Arcobacter most likely did not originate from the crude and 2: 11 phylotypes from both the bacterial and archaeal oil but could become enriched using the components of clone libraries) were related to thermophilic genera that have crude oil in the produced water as a carbon or energy previously been identified from a number of high-temperature source. Additional studies are needed to investigate the petroleum reservoirs world-wide, suggesting that these mi- physiological characteristics of Arcobacter spp. in pro- crobes may be a common component of geothermally heated duced water. subsurface environments (i.e., indigenous bacteria to petro- Firmicutes were found as the second dominant phylum in leum reservoirs) and probably play a role in the geochemical the produced water bacterial community. With over a quarter trophic web of these ecosystems. of the total clones (25.5 %), they were grouped into a phylo- The Hibernia-produced water had a high concentration of type closely related to Thermoanaerobacter sp. (Fig. 2). In the sulfur compounds (726±2 mg/L, Table 1). As mentioned pre- phylogenetic tree, this phylotype formed a cluster adjacent to viously, all the detected phylotypes with large numbers of Thermoanaerobacter mathranii isolated from a hot spring in clones belonged to genera related to sulfur compound- Iceland (Larsen et al. 1997) and another Thermoanaerobacter utilizing microbes (Arcobacter, Thermoanaerobacter, sp. from produced water (Fig. 2). Thermoanaerobacter is Themococcus,andArchaeoglobus), suggesting that these phy- commonly identified in other produced water samples and is lotypes might play a major role in the sulfur cycle (an impor- known as a thermophilic fermentative and organotrophic tant anaerobic process) in the produced water system. sulfur-respiring bacterium. This genus has often been isolated from geographically separated oil reservoirs throughout the Comparison of clone library and DGGE world, which suggests that it might be indigenous to the petroleum reservoir (Cayol et al. 1995; Grassia et al. 1996; Similar to the clone library analyses, the bacterial DGGE Magot et al. 2000), although it cannot be ruled out that it showed a much higher diversity than the archaeal DGGE originates from the water injected into the formation (Dahle (Fig. 3). Although only seven dominant bands were excised et al. 2008). from the bacterial DGGE fingerprints and sequenced for

163 Environ Sci Pollut Res phylogenetic analyses, these seven bands confirmed the iden- measurements comparable with the predicted dilution factors tification of the major phylotypes in the clone libraries (i.e., for discharge models. those related to Thermoanaerobacter, Arcobacter, The q-PCR results also showed that more copies of Alcanivorax). Similarly, the archaeal fingerprints confirmed Thermoanaerobacter were found at 50 m than at 1 m, sug- the identification of members of the two major archaeal clas- gesting that the components of produced water tend to travel ses, Thermococci and Archaeoglobi. These results suggested downward in the water column (Table 3). Yeung et al. (2011a) that a DGGE analysis might be sufficient for monitoring the also found differences in the bacterial community structure in major members in the microbial community in the produced the sediments close to the Thebaud production platform sug- water and seawater in the future. gesting that any influence from the produced water effluent would most likely be seen in the lower part of the water column or in the sediments close to the discharge. The q-PCR Comparison of produced water and seawater bacterial results supported this observation and provided a semi- community structure quantitative method to monitor the zone of influence of the produced water. The DGGE results showed a high similarity in bacterial com- A lower number of copies of Thermoanaerobacter were munity structure for the water column both horizontally and found in the rest of the seawater samples (Table 3), with an vertically, suggesting that there is a spatially stable bacterial average of only around 200 copies in most of the 1-m depth community in the surrounding seawater covering a large re- samples (Table 3). This result suggested that although most of gion (500 m–50 km from the platform). Therefore, any chang- produced water might travel downward in the water column, es in the bacterial community in the water column could po- some components are diluted and transported to other parts of tentially be used as an indicator for the effects from the pro- the water column. Interestingly, the number of copies detected duced water discharge, and this stable bacterial community in at a depth of 50 m was lower at 50 m than at 100 m from the the surrounding seawater could also suggest that any effects platform (Table 3). A possible explanation for this is that the from the produced water were restricted to the region imme- relatively high discharge force of the produced water diately adjacent to the platform (within 500 m). (discharged at a depth of 40 m), may require more distance before it sinks to the 50 m depth. Hibernia seawater q-PCR analysis Nested-PCR analysis The q-PCR results revealed that Thermoanaerobacter was detected in almost all the samples from within 500 m from The nested-PCR method was also used to amplify 2008, showing that components of produced water were de- Thermoanaerobacter in the seawater samples for two pur- tectable in the surrounding seawater within 500 m of the dis- poses: to re-evaluate previous negative results and to confirm charge. The highest concentration of Thermoanaerobacter the positive results from q-PCR analysis. The nested-PCR (and the only value that is truly quantifiable) was found in result was able to confirm all the positive amplifications from seawater sampled at 100 m from the production platform at the q-PCR analysis, and showed weak positive amplification a depth of 50 m, suggesting that most of the produced water from the 1 m sample from 1000 m (Table 3). effluent was transported to that location following discharge. Thermoanaerobacter was not detected in any samples beyond However, only ~26,700 copies/L of Thermoanaerobacter 1000 m using both q-PCR and nested-PCR methods (Ta- were detected at the 100 m location, even though there was bles 3). The nested-PCR results suggested that this method a high concentration of Thermoanaerobacter (~47,750,958 might be able to provide the needed sensitivity to detect the copies/L) in the raw produced water, suggesting that the dilu- low concentrations of Thermoanaerobacter in the surround- tion factor was very high. A 1600-fold dilution factor (i.e., 26, ing seawater and to confirm the findings from other analyses, 700 copies/L÷47,750,958 copies/L×100 %=0.056 %) was such as q-PCR. calculated at the 100-m location from the platform. A modeling study by Somerville et al. (1987) found that even at a 10, 000-m3/day discharge rate, a 100-fold dilution was estimated Conclusions at 50 m from the platform, and a 2800-fold dilution was estimated at 1000 m from the platform. Another modeling study This work provided the first insight into the bacterial and at a low discharge rate (2000 m3/day), Furuholt (1996)esti- archaeal communities in and around the Hibernia offshore mated a 1000-fold dilution would be found at 50 m down- oil and gas production platform. This study characterized both stream from the discharge point. Our findings measured sim- the bacterial and archaeal community structures and chemical ilar dilution factors as those predicted by the modeling studies, compositions of Hibernia-produced water and the indigenous suggesting that using this detection method, we could obtain bacterial community structure and chemical/physicochemical

164 Environ Sci Pollut Res characteristics in the seawater around the platform. All analy- Lysov YP, Perov AN, Mirzabekov AD, Hippe H, Stackebrandt E, ’ ses revealed that the discharge of produced water did not have L Haridon S, Jeanthon C (2003) Radioisotopic, culture-based, and oligonucleotide microchip analyses of thermophilic microbial com- a detectable effect on the bacterial community structure in the munities in a continental high-temperature petroleum reservoir. seawater around the Hibernia production platform. However, Appl Environ Microbiol 69:6143–6151 the results also revealed that the produced water has a rich Brakstad OG, Lødeng AG (2005) Microbial diversity during biodegrada- – microbial diversity and has unique chemical properties, but tion of crude oil in seawater from the North Sea. Microb Ecol 49:94 103 these specific characteristics appeared to be below detection Brakstad OG, Nonstad I, Faksness LG, Brandvik PJ (2008) Responses of limits in the surrounding water. Hence, signature microorgan- microbial communities in Arctic sea ice after contamination by isms from produced water could be used as targets to monitor crude petroleum oil. Microb Ecol 55:540–552 the dispersion of produced water in the surrounding ocean Brooks JM, Estes EL, Wiesenburg DA, Schwab CR, Abdel-Rheim HA (1980) Investigations of surficial sediments, suspended particulates, using methods like q-PCR and nested-PCR. Both q-PCR and volatile hydrocarbons at Buccaneer gas and oil field. In: Jackson and nested-PCR results revealed that Thermoanaerobacter WB, Wilkens EP (eds) Environmental assessment of the Buccaneer spp. were present in the surrounding seawater near the dis- gas and oil field in the Northwestern Gulf of Mexico, 1978–1980, charge but were not detected beyond 1000 m from the Hiber- vol 1, NOAA Technical Memorandum NMFS-SEFC-47. National Oceanic and Atmospheric Administration, National Marine nia production platform, and as such, could be potential target Fisheries Service, Galveston, p 98 microorganisms for tracking produced water discharges. Brooks SJ, Harman C, Grung M, Farmen E, Ruus A, Vingen S, Godal BF, Baršiene J, Andreikenaite L, Skarphéoinsdóttir H, Liewenborg B, Acknowledgments The authors thank Jennifer Mason from the Centre Sundt RC (2011) Water column monitoring of the biological effects of Offshore Oil, Gas and Energy Research for analytical and technical of produced water from the ekofisk offshore oil installation from support. 2006 to 2009. J Toxicol Environ Health Part A 74:582–604 Cayol JL, Ollivier B, Patel BKC, Ravot G, Magot M, Ageron E, Grimont Compliance with ethical standards This research represents original PAD, Garcia JL (1995) Description of Thermoanaerobacter brockii data that has not been submitted elsewhere for publication, that all ac- subsp lactiethylicus subsp nov, isolated from a deep subsurface knowledgements have been made, and that all authors have contributed French oil-well, a proposal to reclassify Thermoanaerobacter finii significantly to the work. There are no real or perceived conflicts of as Thermoanaerobacter brockii subsp finnii comb nov, and an interest. emended description of Thermoanaerobacter brockii. Int J Syst Bacteriol 45:783–789 Chao A (1984) Non-parametric estimation of the number of classes in a population. Scand J Stat 11:265–270 References Collins AG (1975) Geochemistry of oilfield waters. Elsevier Scientific Publishers, New York Colwell RK (2005) EstimateS: statistical estimation of species richness Abrahamson A, Brandt I, Brunström B, Sundt R, Jørgensen E (2008) and shared species from samples. Version 7.5. User’s Guide and Monitoring contaminants from oil production at sea by measuring application published at: http://purl.oclc.org/estimates gill EROD activity in Atlantic cod (Gadus morhua). Environ Pollut Conmy RN, Coble PG, Farr J, Wood AM, Lee K, Pegau S, Walsh I, Koch – 153:169 175 C, Abercrombie MI, Miles MS, Lewis MR, Ryan R, Robinson B, Adams J, Bornstein JA, Munno K, Hollebone B, King T, Brown RS, King T, Kelble C, Lacoste J (2014) Submersible optical sensors Hodson PV (2014) Identification of compounds in heavy fuel oil exposed to chemically dispersed crude oil: wave tank simulations that are chronically toxic to rainbow trout embryos by effects-driven for improved oil spill monitoring. Environ Sci Technol 48:1803– chemical. Environ Toxicol Chem. doi:10.1002/etc.2497 1810 Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local Dahle H, Garshol F, Madsen M, Birkeland NK (2008) Microbial com- – alignment search tool. J Mol Biol 215:403 410 munity structure analysis of produced water from a high- Amann RI, Ludwig W, Schleifer K-H (1995) Phylogenetic identification temperature North Sea oil-field. Antonie Van Leeuwenhoek 93: and in situ detection of individual microbial cells without cultiva- 37–49 – tion. Microbiol Rev 59:143 169 Davydova-Charakhch’yan IA, Mileeva AN, Mityushina LL, Belyaev SS Armstrong FAJ, Sterns CR, Strickland JDH (1967) The measurement of (1992) Acetogenic bacteria from oil fields of Tataria and Western upwelling and subsequent biological processes by means of the Siberia (Engl. Trans.). Microbiology 61:306–315 technicon autoanlyzer and associated equipment. Deep-Sea Res Flynn SA, Butler EJ, Vance I (1995) Produced water composition, toxic- 14:381–389 ity and fate: a review of recent BP North Sea studies. In: Reed M, Azam F, Malfatti F (2007) Microbial structuring of marine ecosystems. Johnsen S (eds) Produced water 2. Plenum Press, New York, pp 69– Nat Rev Microbiol 5:782–791 80 Banat IM (1995) Biosurfactants production and possible uses in microbial Fortin N, Fulthorpe RR, Allen DG, Greer CW (1998) Molecular analysis enhanced oil recovery and oil pollution remediation: a review. of bacterial isolates and total community DNA from kraft pulp mill Bioresour Technol 51:1–12 effluent treatment systems. Can J Microbiol 44:537–546 Banat IM, Makkar RS, Cameotra SS (2000) Potential commercial appli- Furuholt E (1996) Environmental effects of discharge and reinjection of cations of microbial surfactants. Appl Microbiol Biotechnol 53: produced water. In: Reed M, Johnsen S (eds) Produced water 2. 495–508 Environmental issues and mitigation technologies. Plenum Press, Beeder J, Nilsen RK, Rosnes JT, Torsvik T, Lien T (1994) Archaeoglobus New York, pp 275–288 fulgidus isolated from hot North-Sea-oil field waters. Appl Environ Galluzzi L, Penna A, Bertozzini E, Vila M, Garces E, Magnani M (2004) Microbiol 60:1227–1231 Development of a real-time PCR assay for rapid detection and quan- Bonch-Osmolovskaya EA, Miroshnichenko ML, Lebedinsky AV, tification of Alexandrium minutum (a dinoflagellate). Appl Environ Chernyh NA, Nazina TN, Ivoilov VS, Belyaev SS, Boulygina ES, Microbiol 70:1199–1206

165 Environ Sci Pollut Res

Gauthier MJ, Lafay B, Christen R, Fernandez L, Acquaviva M, Bonin P, L’Haridon S, Miroshnichenko ML, Hippe H, Fardeau ML, Bonch- Bertrand JC (1992) Marinobacter hydrocarbonoclasticus gen. nov., Osmolovskaya EA, Stackebrandt E, Jeanthon C (2002) Petrotoga sp. nov., a new, extremely halotolerant, hydrocarbon-degrading ma- olearia sp. nov., and Petrotoga sibirica sp. nov., two thermophilic rine bacterium. Int J Syst Bacteriol 42:568–576 bacteria isolated from a continental petroleum reservoir in Western Grabowski A, Nercessian O, Fayolle F, Blanchet D, Jeanthon C (2005) Siberia. Int J Syst Evol Microbiol 52:1715–1722 Microbial diversity in production waters of a low-temperature Large R (1990) Characterization of produced water, phase 1: literature biodegraded oil reservoir. FEMS Microbiol Ecol 54:427–443 survey. Report to Conoco (UK) Ltd., Aberdeen, Scotland Grassia GC, McLean KM, Glenat P, Bauld J, Sheehy A (1996) A sys- Larsen L, Nielsen P, Ahring BK (1997) Thermoanaerobacter mathranii tematic survey for thermophilic fermentative bacteria and archaea in sp. nov., an ethanol-producing, extremely thermophilic anaerobic high temperature petroleum reservoirs. FEMS Microbiol Ecol 21: bacterium from a hot spring in Iceland. Arch Microbiol 168:114– 47–58 119 Guo CL, Zhou HW, Wong YS, Tam NF (2005) Isolation of PAH- Le Petit J, Bertrand JC, N’Guyen MH, Tagger S (1975) On the taxonomy degrading bacteria from mangrove sediments and their biodegrada- and physiology of bacteria utilizing hydrocarbons in the sea. Ann tion potential. Mar Pollut Bull 51:1054–1061 Microbiol (Paris) 126:367–380 Hahn MW (2006) The microbial diversity of inland waters. Curr Opin Lee K, Cobanli S, Robinson R, and Wohlgeschaffen G (2011) Biotechnol 17:256–261 Application of microbiological methods to assess the potential im- Han P, Zheng L, Cui ZS, Guo XC, Tian L (2009) Isolation, identification pact of produced water discharges. In: Lee K, Neff J (eds) Produced and diversity analysis of petroleum-degrading bacteria in Shengli water environmental risks and advances in mitigation technologies. Oil Field wetland soil. Ying Yong Sheng Tai Xue Bao 20:1202– pp. 353–373 1208 Li H, Yang SZ, Mu BZ, Rong ZF, Zhang J (2006) Molecular analysis of Hansen MC, Tolker-Neilson T, Givskov M, Molin S (1998) Biased 16S the bacterial community in a continental high-temperature and rDNA PCR amplification caused by interference from DNA water-flooded petroleum reservoir. FEMS Microbiol Lett 257:92– flanking the template region. FEMS Microbiol Ecol 26:141–149 98 Harman C, Thomas KV, Tollefsen KE, Meier S, Boyum O, Grung M Li H, Yang SZ, Mu BZ, Rong ZF, Zhang J (2007) Molecular phylogenetic (2009) Monitoring the freely dissolved concentrations of polycyclic diversity of the microbial community associated with a high- aromatic hydrocarbons (PAH) and alkylphenols (AP) around a temperature petroleum reservoir at an offshore oilfield. FEMS Norwegian oil platform by holistic passive sampling. Mar Pollut Microbiol Ecol 60:74–84 Bull 58:1671–1679 Liesack W, Weyland H, Stackebrandt E (1991) Potential risks of gene Hayek LC, Buzas MA (1996) Surveying natural populations. Columbia amplification by PCR as determined by 16S rDNA analysis of a University Press, New York mixed culture of strict barophilic bacteria. Microb Ecol 21:191–198 Hedlund BP, Staley JT (2006) Isolation and characterization of Liu YJ, Chen YP, Jin PK, Wang XC (2009) Bacterial communities in a Pseudoalteromonas strains with divergent polycyclic aromatic hy- crude oil gathering and transferring system (China). Anaerobe 15: drocarbon catabolic properties. Environ Microbiol 8:178–182 214–218 Hylland K, Tollefsen K, Ruus A, Jonsson G, Sundt RC, Sanni S, Røe Magot M, Caumette P, Desperrier JM, Matheron R, Dauga C, Grimont F, Utvik TI, Johnsen S, Nilssen I, Pinturier L, Balk L, Barsiene J, Carreau L (1992) Desulfovibrio longus sp-Nov, a sulfate-reducing Marigòmez I, Feist SW, Børseth JF (2008) Water column monitor- bacterium isolated from an oil producing well. Int J Syst Bacteriol ing near oil installations in the North Sea 2001–2004. Mar Pollut 42:398–403 Bull 56:414–429 Magot M, Ollivier B, Patel BKC (2000) Microbiology of petroleum res- Iwabuchi N, Sunairi M, Urai M, Itoh C, Anzai H, Nakajima M, Harayama ervoirs. Antonie van Leeuwenhoek 77:103–116 S (2002) Extracellular polysaccharides of Rhodococcus Magurran AE (1988) Ecological diversity and its measurement. Princeton rhodochrous S-2 stimulate the degradation of aromatic components University Press, Princeton in crude oil by indigenous marine bacteria. Appl Environ Microbiol Magurran AE (2004) Measuring biological diversity. Blackwell 68:2337–2343 Maidak BL, Cole JR, Lilburn TG, Parker CT Jr, Saxman PR, Farris RJ, Jha AN (2004) Review. Genotoxicological studies in aquatic organisms: Garrity GM, Olsen GJ, Schmidt TM, Tiedje JM (2001) The RDP-II an overview. Mutat Res 552:1–17 (ribosomal database project). Nucleic Acids Res 29:173–174 Jha AN (2008) Ecotoxicological applications and significance of the Martinez-Murcia AJ, Acinas SG, Rodriguez-Valera F (1995) Evaluation comet assay. Mutagenesis 23:207–221 of prokaryotic diversity by restrictase digestion of 16S rDNA direct- Kaster KM, Bonaunet K, Berland H, Kjeilen-Eilertsen G, Brakstad OG ly amplified from hypersaline environments. FEMS Microbiol Ecol (2009) Characterisation of culture-independent and -dependent mi- 17:247–256 crobial communities in a high-temperature offshore chalk petroleum Meyer A, Moser R, Neef A, Stahl U, Kampfer P (1999) Differential reservoir. Antonie Van Leeuwenhoek 96:423–439 detection of key enzymes of polyaromatic-hydrocarbon-degrading Keller M, Braun F-J, Dirmeier R, Hafenbradl D, Burggraf S, Rachel R, bacteria using PCR nad gene probes. Microbiology 145:1731–1741 Stetter KO (1995) Thermococcus alcaliphilus sp. nov., a new hyper- Murphy J, Riley JP (1962) A modified single solution method for the thermophilic archaeum growing on polysulfide at alkaline pH. Arch determination of phosphate in natural waters. Anal Chim Acta 27:30 Microbiol 164:390–395 Nazina TM, Ivanova AE, Borzenkov IA, Belyaev SS, Ivanov MV (1995) Kerouel R, Aminot A (1997) Fluorometric determination of ammonia in Occurrence and geochemical activity of microorganisms in high- sea and estuarine waters by direct segmented flow analysis. Mar temperature, water-flooded oil fields of Kazakhstan and Western Chem 57:265–275 Siberia. Geomicrobiol J 13:181–192 Kobayashi H, Endo K, Sakata S, Mayumi D, Kawaguchi H, Ikarashi M, Neff JM (1987) Biological effects of drilling fluids, drill cuttings and Miyagawa Y, Maeda H, Sato K (2012) Phylogenetic diversity of produced waters. In: Boesch DF, Rabalais NN (eds) Long-term ef- microbial communities associated with the crude-oil, large- fects of offshore oil and gas development. Elsevier Applied Science insolble-particle and formation-water components of the reservoir Publishers, London, pp 469–538 fluid from a non-flooded high-temperature petroleum reservoir. J Neff JM (2002) Bioaccumulation in marine organisms. effects of contam- Biosci Bioeng 113:204–210 inants from oil well produced water. Elsevier Science Publishers, Krebs C (1989) Ecological methodology. Harper and Row, New York Amsterdam, 452 pp

166 Environ Sci Pollut Res

Neuner A, Jannasch HW, Belkin S, Stetter KO (1990) Thermococcus Simpson EH (1949) Measurement of diversity. Nature 163:688 litoralis sp. nov.: a new species of extremely thermophilic marine Somerville HJ, Bennett D, Davenport JN, Hold MS, Lynes A, Mahiew A, archaebacteria. Arch Microbiol 153:205–207 McCourt B, Parker JG, Stephenson RR, Watkinson RJ, Wilkinson Orphan VJ, Taylor LT, Hafenbradl D, DeLong EF (2000) Culture- TG (1987) Environmental effect of produced water from North Sea dependent and culture-independent characterization of microbial as- oil operations. Mar Pollut Bull 18:49–558 semblages associated with high-temperature petroleum reservoirs. Stagg RM, McIntosh A (1996) Hydrocarbon concentrations in the north- Appl Environ Microbiol 66:700–711 ern North Sea and effects on fish larvae. Sci Total Environ 186:189– Orphan VJ, Goffredi SK, DeLong EF, Boles JR (2003) Geochemical 201 influence on diversity and microbial processes in high temperature Stetter KO (1988) Archaeoglobus fulgidus gen-nov, sp-nov—a new taxon oil reservoirs. Geomicrobiol J 20:295–311 of extremely thermophilic archaebacteria. Syst Appl Microbiol 10: Paine MD, Leggett WC, McRuer JK, Frank KT (1992) Effects of 172–173 Hibernia crude oil on capelin (Mallotus villosus) embryos and lar- Strickland JDH, Parsons TR (1973) A practical handbook of seawater vae. Mar Environ Res 33:159–187 analysis. Fish Res Board Can Bull 1:155–164 Piubeli F, Grossman MJ, Fantinatti-Garboggini F, Durrant LR (2014) Suzuki MT, Taylor LT, DeLong EF (2000) Quantitative analysis of small- Phylogenetic analysis of the microbial community in hypersaline subunit rRNA genes in mixed microbial populations via 5′-nuclease petroleum produced water from the Campos Basin. Environ Sci assays. Appl Environ Microbiol 66:4605–4614 Pollut Res 21:12006–12016 Terrens GW, Tait RD (1996) Monitoring ocean concentrations of aromat- Prince RC, Bare RE, Garrett RM, Grossman MJ, Haith CE, Keim LG, ic hydrocarbons from produced formation water discharges to Bass Lee K, Holtom GJ, Lambert P, Sergy GA, Owens EH, Guénette CC Strait, Australia. SPE 36033. In: Proceedings of the International (1999) Bioremediation of a marine oil spill in the Arctic. In: Conference on Health, Safety & Environment. Society of Alleman BC, Leeson A (eds) In-situ bioremediation of petroleum Petroleum Engineers, Richardson, pp 739–747 hydrocarbon and other organic compounds. Battelle, Columbus, pp Tibbets PJC, Buchanan IT, Gawel LJ, Large R (1992) A comprehensive 227–232 determination of produced water composition. In: Ray JP, Englhardt Rabalais NN, McKee BA, Reed DJ, Means JC (1991) Fate and effects of FR (eds) Produced water. Plenum Press, New York, pp 97–112 nearshore discharges of OCS produced waters. Vol. 1: executive Voordouw G, Armstrong SM, Reimer MF, Fouts B, Telang AJ, Shen Y, summary. Vol. 2: Technical Report. Vol. 3. Appendices OCS Gevertz D (1996) Characterization of 16S rRNA genes from oil field Studies MMS 91-004. MMS 91-005, and MMS 91-006. U.S. microbial communities indicates the presence of a variety of sulfate- Dept. of the Interior, Minerals Management Service, Gulf of reducing, fermentative, and sulfide-oxidizing bacteria. Appl Mexico OCS Regional Office, New Orleans, LA Environ Microbiol 62:1623–1629 Ren H-Y, Zhang X-J, Song Z-Y, Rupert W, Gao G-J, Guo S-X, Zhao L-P Wawrik B, Paul JH, Tabita FR (2002) Real-time PCR quantification of (2011) Comparison of microbial community compositions of injec- RbcL (ribulose-1,5-bisphosphate carboxylase/oxygenase) mRNA in tion and production well samples in a long-term water-flooded pe- diatoms and pelagophytes. Appl Environ Microbiol 68:3771–3779 troleum reservoir. PLoS One 6:e23258 Yakimov MM, Golyshin PN, Lang S, Moore ERB, Abraham W-R, Reysenbach A-L, Pace NR (1995) In: Robb FT, Place AR (eds) Archaea: Lunsdorf H, Timmis KM (1998) Alcanivorax borkumensis gen. a laboratory manual—thermophiles. Cold Spring Harbor Press, nov., sp. nov., a new, hydrocarbon-degrading and surfactant- Cold Spring Harbor, pp 101–107 producing marine bacterium. Int J Syst Bacteriol 48:339–348 Rittenhouse G, Fulron RB, Grabowski RJ, Bernard JL (1969) Minor Yeung CW, Lee K, Whyte LG, Greer CW (2010) Microbial community elements in oil field waters. Chem Geol 4:189–209 characterization of the gully: a marine protected area. Can J Røe Utvik TI, Durell GS, Johnsen S (1998) Determining produced water Microbiol 56:421–431 originating polycyclic aromatic hydrocarbons in North Sea waters: Yeung CW, Law TG, Milligan K, Lee K, Whyte LG, Greer CW (2011a) comparison of sampling techniques. Mar Pollut Bull 38:977–989 Analysis of bacterial diversity and metals in produced water, seawa- Romine MF, Stillwell LC, Wong KK, Thurston SJ, Sisk EC, Sensen C, ter and sediments from an offshore oil and gas production platform. Gaasterland T, Fredrickson JK, Saffer JD (1999) Complete sequence Mar Pollut Bull 62:2095–2105 of a 184-kilobase catabolic plasmid from Sphingomonas Yeung CW, Lee K, Greer CW (2011b) Microbial community character- aromaticivorans F199. J Bacteriol 181:1585–1602 ization of produced water from the Hibernia oil production platform. Rozen S, Skaletsky HJ (2000) Primer3 on the WWW for general users In: Lee K, Neff J (eds) Produced water: environmental risks and and for biologist programmers. In: Misener S, Krawetz SA (eds.). advances in mitigation technologies. Springer, New York Bioinformatics methods and protocols: methods in molecular biol- Yeung CW, Woo M, Lee K, Greer CW (2011c) Characterization of the ogy. Humana Press Inc., Totowa (NJ). pp 365–386. Source code bacterial community structure of Sydney Tar Ponds sediment. Can J available at http://fokker.wi.mit.edu/primer3/ Microbiol 57:493–503 Rybakovas A, Baršiene J, Lang T (2009) Environmental genotoxicity and Zhu F, Massana R, Not F, Marie D, Vaulot D (2005) Mapping of cytotoxicity in the offshore zones of the Baltic and the North Seas. picoeucaryotes in marine ecosystems with quantitative PCR of the Mar Environ Res 68:246–256 18S rRNA gene. FEMS Microbiol Ecol 52:79–92

167 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE A numerical model to simulate the droplet formation process resulting from the release of diluted bitumen products in marine environment

Lin Zhao1, Jagadish Torlapati1, Thomas King2, Brian Robinson2, Michel C. Boufadel1, and Kenneth Lee3

1Center for Natural Resources Development and Protection, Department of Civil and Environmental Engineering, the New Jersey Institute of Technology, Newark, NJ, USA. 07102 [email protected]; http://nrdp.njit.edu 2Bedford Institute of Oceanography, Department of Fisheries and Oceans, Dartmouth, Canada. 3Flagship of Wealth from Oceans, Commonwealth Scientific and Industry Research Organization (CSIRO), Perth, Australia.

ABSTRACT 300315:

A numerical model that simulates the dispersion of oil due to the action of waves in the marine environment is presented. Model validations were performed in association with the wave tank experiments conducted in the Department of Fisheries and Oceans (DFO) Canada. Two dilbit products were considered: Access Western Blend and Cold Lake Blend. The oil droplet size distribution in the subsurface water column obtained from the experimental observations was reproduced using the droplet formation model. Special consideration was made for the simulation of wave effects on surface oil spills. Modeling results show the successful use of droplet formation model in the simulation of oil spills due to wave actions.

INTRODUCTION:

Oil spills in the open ocean have received great attention in recent years because of the Deepwater Horizon oil spill (BP, 2010). The spilled oil on the water surface can be transported by natural dispersion processes, and wave actions may also lead to the formation of small droplets in the water column. The droplet size distribution (DSD) is very important for the study of the fate and transport of the spilled oil. Small oil droplets will result in increased area of the oil-water interface, which enhances the dissolution of hydrocarbon compounds in the water column and accelerates oil biodegradation (Reddy et al., 2012; Nicol et al., 1994; Geng et al., 2013).

Population balance equation is the most widely used method for the simulation of droplet formation processes (Bandara and Yapa, 2011; Prince and Blanch, 1990; Tsouris and Tavlarides, 1994). This numerical approach takes into account various mechanisms causing breakup and coalescence of droplets. One of the advantages of the population balance model is that it provides transient size distributions. In the peer reviewed investigations, only few studies considered oil viscosity effects in droplet breakage, and these occurred under steady state conditions (Coulaloglou and Tavlarides, 1977; Wang and Calabrese, 1986). Also, most studies focus on simulations in stirred-tank experiments, and we are not aware of any papers dealing with the DSD evolution caused by waves.

449

168 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE These limitations make our investigation of dilbit products in a wave tank very timely, as the dilbit has typically high viscosity. Dilbit is short for “diluted bitumen”, and is obtained by blending various high viscosity oils with diluents (typically condensate) to allow transportation of the blend in pipes. Yang et al. (2011) pointed out that the chemical features of bitumen in oil sands are clearly distinguishable from those of most conventional crude oils and the chemical characteristics of diluted crude bitumen are altered significantly due to blending with diluents. Therefore, the dispersive behavior of dilbit products may be different from those of conventional crude oil, and we are only beginning to study and understand its dispersion processes.

Therefore, the objective of this study is to introduce the model VDROP, a physically-based droplet evolution model with the capability of modeling both low- and high-viscosity oils. The model is used to simulate the droplet size distribution of two dilbit products under breaking wave conditions in wave tank experiments which were conducted in the Bedford Institute of Oceanography, Nova Scotia, Canada.

METHODOLOGIES:

Model Development: The droplet formation model, VDROP, which uses the population balance equation for a discrete droplet size system and considers both viscosity and interfacial tension of drops as resistance forces, is used to account for the droplet breakage and coalescence mechanisms. Detailed model development and validations can be found in Zhao et al. (2014). The methodologies of the VDROP model are summarized below.

Considering a liquid-liquid dispersive system, the rate of the generation of droplets with diameter of di can be expressed as a result of birth “B” and death “D”:

 tdn ),( i  tdDtdB ),(),( (1) t i i

3 where n is number concentration (number/m ) of droplets of diameter di (m) at a given time t (s). For a discrete droplet size system, the birth and death of drops with size di are given as

n n n i   jjji tdndgddtdB ),()(),(),(   jkj k tdntdndd ),(),(),( (2) ij  1 j1k  1  vvv ikj

n i  ii  i  jji tdnddtdntdndgtdD ),(),(),(),()(),( (3) j1 where β(di,dj) is the breakage probability density function for the creation of droplet of diameter di due to breakage of droplets of (a larger) diameter dj (dimensionless), and g(dj) is the breakage 3 frequency of droplets of diameter dj (/s), Г(dk,dj) is the coalescence rate (m /s).The first term of birth “B” represents births of droplets of size di resulting from the breakup of droplets of size dj, while the second term represents the birth of droplets of size di as a result of coalescence events 450

169 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE occurring between droplets of smaller sizes dk and dj to form drops with the size of di; The first term of death “D” represents deaths of droplets of size di due to breakup into smaller droplets, while the second term represents deaths of droplets of size di due to the coalescence of droplets of size di with droplets of all sizes (including other droplets of size di themselves) to form larger droplets.

Droplet breakup There are a number of mechanisms for droplet breakage. The major mechanism is the bombardment of droplets by turbulent eddies (Coulaloglou and Tavlarides, 1977; Luo and Svendsen, 1996; Tsouris and Tavlarides, 1994; Walter and Blanch, 1986). The breakage rate of a droplet can be written as the product of the collision frequency (total number of collisions per time) between a droplet and all the surrounding eddies, which is estimated based on the analogy of kinetic collisions of ideal gas molecules first developed by Kennard (1938); and breakage efficiency, BE, which presents the probable occurrence of breakage events due to the collisions, since not all collisions induce droplet breakage. In addition, the model VDROP contains a formulation that accounts for resistance forces due to the viscosity of the oil droplet. With this concept in the model development, the breakage rate g(di) can thus be expressed as (Tsouris and Tavlarides, 1994):

  2/122 bi  ed uuSKdg de BE ei tdd ),,()()( dne (4) ne where Kb is the adjustable parameter for droplet breakup, Sed represents the cross section area of 2 eddy-droplet (m ), ue is the average turbulent velocity of an eddy (m/s), ud is droplet average 3 velocity (m/s), ne is the number concentration of eddies (number of eddies/m ), and size de is the diameter of the eddies (m). In the inertial subrange of the energy spectrum (Kolmogorov, 1941), the turbulent velocity of an eddy, ue, and drop velocity, ud, can be expressed as (Azbel, 1981):

3/1 e  .2 27 du e )( (5)

3/1 d  .1 03 du i )( (6) where  is the energy dissipation rate (watt/kg). The velocities are obtained as averages for both the eddy and droplet velocities. Detailed formulations for each term are described in Zhao et al. (2014).

The parameter Kb (dimensionless) is a system-dependent term that would need to be obtained by calibration to experimental data. It is highly desirable that this parameter does not change much when changes are made within a system; for example if Kb is equal to 3.0 based on breaking waves of mixing energy of 1.5 watt/hour, it is desirable that its value does not change if the mixing energy due to waves increases to 5 watts/hour. Time-dependent viscous effects BE

451

170 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE The breakage efficiency term (dimensionless), BE, can be written as:  1   EE  BE exp   vc  (7)  c1  e 

2 2 where Ec is the average excess of surface energy (kg m /s ), needed to form a pair of equal-size daughter droplets or a small and large droplets, (note that this term also known as formation 2 2 energy); Ev is the resistance energy due to viscous forces within the droplet (kg m /s ), ; e is the 2 2 energy of the turbulent eddy (kg m /s ), and c1 is an empirical constant equal to 1.3 (Tsouris and Tavlarides, 1994).

Most of the previous investigators considered the viscous resistance forces in a steady-state DSD (Calabrese et al., 1986; Coulaloglou and Tavlarides, 1977; Wang and Calabrese, 1986). However, when viscosity is the primary resisting force to droplets breakage, the viscous breakup process may be time-dependent. Droplets break by elongation of the droplet until reaching an unstable length, typically 2 to 3 times the initial diameter (Taylor, 1934). As turbulence is not homogeneous (Frisch and Parisi, 1985; Frisch et al., 1978), the volume of high intensity mixing is usually small relative to the whole domain, and thus, only time would ensure that a droplet is subjected to a high mixing intensity for a sufficiently long duration. However, it is not possible to trace the movement and shape changes of each droplet (millions of droplets could exist in a small stirred vessel). Therefore, using physical arguments based on Hinze (1955) and Calabrese et al. (1986), we developed a time –dependent viscous force energy expressed as:

  t       3/73/1  c  v  aE exp1   d di  (8)   T  6 d  where a is a constant calibrated to be 1.1, t is simulation time (s), T is the probable mean lifetime of the system calibrated to be 8400 s, which may present the average time to reach steady state condition for high viscosity oils, d is the viscosity of dispersed phase (Pa·s), c and d are the fluid density of continuous and dispersed phase (kg/m3), respectively.

Droplet coalescence The coalescence rate Γ(di,dj), can be described as the product of collision frequency and coalescence efficiency. Collision events of two droplets are considered herein as due only to turbulence. Other mechanisms, such as buoyancy and lateral shear (Prince and Blanch, 1990) will be considered in future works.

When two droplets collide, they either bounce off or coalesce to form a larger droplet. For the process of coalescence to occur, the two droplets must be in contact with each other long enough to rupture the film between them. The time to thin the film until it reaches a critical film thickness is called the coalescence time. Theoretically, for coalescence to occur, the contact time between two droplets should be greater than the coalescence time. Therefore, the coalescence rate is expressed as:

2/122 tij cji ij  uuSKdd ji  )exp()(),( (9)  ij

452

171 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE where Kc is the adjustable parameter for the simulation of droplet coalescence (dimensionless), 2 Sij represents the cross section area (m ) of the collided droplets with sizes di and dj, ui and uj are the velocities (m/s) of droplets di and dj, respectively, tij is the coalescence time (s), and τij is the contact time (s) between the two droplets di and dj. The contact time τij is given by Prince and Blanch (1990) as:

3/2  1   22   5.0     3/2    dd ji  rij       (10) ij  3/1  3/1 where rij is the equivalent radius (m).

Numerical implementation Eq. 1 is integrated in time using first-order Euler method. As shown in Eq. 1, the function solves the droplet breakup and coalescence processes; however, it does not ensure the conservation of mass (or volume, if the fluids under consideration are incompressible). To solve the mass balance problem in Eq. 1, the following method was used. As shown in Figure 1, considering one droplet of size d0 breaks up to form two droplets of sizes d1 and d2, the droplet of size d0 becomes the parent droplet (this droplet 0 will be considered dead in the system after it breaks up into two smaller ones), while the droplet of size d1 is a daughter droplet, and the droplet of size d2 is the complementary daughter droplet formed from the remaining volume (volume v0– volume v1). This complementary daughter droplets’ actual diameters (d2) may lie in different bin sizes, and the volume of the complementary daughter droplet is lost if it is not accounted for in the model. Therefore, to ensure mass conservation, we find the bin location of the complementary daughter droplet’s diameter (i.e., the maximum and minimum sizes of the droplet bin wherein the diameter lies as shown in Figure 1) and the remaining volume is interpolated between the two droplet sizes. The droplet is assumed spherical throughout the simulation when calculating volume from droplet size (diameter or equivalent diameter). A similar approach is used for coalescence where the volume of the droplet formed from the coalescence of two droplets is distributed into the nearest bin sizes by interpolation. The mass is always conserved using such an approach as the volumes are explicitly accounted for within each droplet bin size.

Experimental setup The experiments were conducted at a wave tank facility located at Bedford Institute of Oceanography (BIO) (Dartmouth, Nova Scotia, Canada) to study the fate and transport of dilbit products following a hypothetical oil spill in a marine environment. The wave tank facility measures 32 m long, 0.6 m wide, and 2 m high, with an average water depth of 1.5 m. A paddle is situated at 3 m from the front wall of the tank to generate different patterns of waves. For this study, breaking waves were generated using the dispersive focusing technique (Rapp and Melville, 1990; see also Wickley-Olsen et al., 2007, for a related approach), with the resulting breaker height being around 0.40 m. Unfortunately, at the time of preparation of this paper, the water velocity profile was not measured during the experiments to compute the energy dissipation rate. Based on studies of Venosa et al. (2005) and Li et al. (2008), for a wave height of 0.12 m, the high energy dissipation rate in the mixing zone of the wave tank was computed as

453

172 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE 0.5 watt/kg. Here, the high energy dissipation rate was estimated at 1.0 to 3.0 watt/kg based on the breaker height of about 0.4 m.

Two types of dilbit products were used in this study – Access Western Blend (AWB), and Cold Lake Blend (CLB). Prior to use in the experiments, both oils were artificially weathered (8.8% for AWB, 6.2% for CLB). Properties of these two oils are shown in Table 1. The wave tank was filled with natural seawater from the Bedford Basin of Halifax harbor with an average temperature of 8 ºC and salinity of 28 ppt. For each experiment, the oil was poured into a containment ring on the surface of the water. Three replicate experiments for each oil were conducted to obtain average droplet size distributions.

Oil droplet size distribution dispersed in the water column was measured by laser diffraction using two LISST (Laser in-Situ Scattering and Transmissometry) particle-size analyzers situated 1.2 m and 12 m downstream of the oil release at a depth of 45 cm, respectively. The particle size distributions were subdivided into 32 particle size intervals located logarithmically from 2.5 to 500 m in diameter. In this study, the highest total droplet concentration were in the range of 2 L/L – 90 L/L, which indicated that the holdup (volume fraction of the dispersed oil in the system) was very small with an average of 30 × 10-6.

For each oil, three replicate experiments were conducted. Table 2 shows the comparison of d50 value to that calculated from the averaged DSD from the three replicates. The variations of AWB replicates are less than 2%, while variations of CLB replicates are relatively large with 18% for replicate 1, but less than 7% for the other two. The DSDs of the three replicates and the averaged one are presented in the Results Analysis Section below. Variability in the three replicates from the wave tank experiments could be caused by the uncontrolled environmental variables during the experiments (the wave tank facility at BIO was constructed outside, close to the water), such as wind speed, air and water temperature, salinity etc.

RESULTS ANALYSIS:

There are many unknown factors affecting the dispersion of surface oil by waves. These include the energy and duration of the impact from the breaker, the resurfacing of larger droplets, and transport of droplets of various sizes to different depths. In addition, the wave breaker is a transient process, which means after one wave passes, there is a calm period, until the next breaker occurs. Therefore, to simulate wave forces on dispersed oil, the concept of “passes” is used which is described below.

The effect of waves is simulated by varying the energy dissipation rate every 2 min (the period of generating one breaking wave in the experiment was about 2 min). Based on experimental studies in stirred tanks, droplets break only in the high energy zone, therefore, the estimated highest energy dissipation rate of 1.5 W/kg is used for the simulation, and a 6-s duration is assumed to work on the oil and to cause droplet breakage to occur in one wave pass. The selection of 6-s duration is discussed further in the Discussion section. The model is configured as: during the first 6s the energy dissipation rate is set at 1.5 W/kg, then after the initial 6s (6s

454

173 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE collected between 2 min to 20 min, the total simulation time is 20 min. The adjustable parameter for droplet breakups, Kb in Eq. 4, was set to 2.5 for all simulations based on the system setup. The time in calculation of viscosity resistance forces (Eq.8) is updated only when the energy dissipation rate is at or above the design value of 1.5 watt/kg. The viscosity function is meant to account for the heterogeneity of turbulence and that a viscous droplet would need sufficient time in the system to encounter energetic eddies that would break it. For this reason, when there is no mixing, the time in the viscosity function is stopped until the mixing energy reaches the “design” value of 1.5 W/kg. During this time, it is assumed that the droplets will stay in the mixing zone and wait for another breaking wave. As noticed that during the experiment, after the breaking wave passes, the water does not go quiescent immediately and there are still mild waves passing by. Therefore, we assumed that the stress caused by the last breaking wave accumulates to the next breaking wave. Because the holdup of dispersed phase is very small (average value of 30 × 10-6), the droplet coalescence events are negligible in this study. The initial droplet size distribution (DSD) consisted of a seed diameter in the largest size bin (461 microns). Also, it is assumed that all the droplets of size less than 100 microns are lost after 60 seconds, which is done to represent the entrainment in the water column by waves (Delvigne and Sweeney, 1988).

Since the current modeling study does not include the advection process of the droplet movements along the tank, only results at 1.2 m downstream distance from the oil release point are compared. The AWB dilbit product has relatively higher viscosity (348 cp) than CLB (175 cp), implying that the AWB dilbit product would be more difficult to break into small droplets than CLB. This is depicted by both the experimental and modeling results as shown in Figures 2 and 3.With the same wave energy, for AWB (Figure 2), most of the oil in the water column consisted of droplets larger than 331.1 m in diameter, while a relatively larger amount of smaller droplets were dispersed in the water column for the CLB product (Figure 3).

Plots of both volume probability density function (Figure 2a) and cumulative volume fraction (Figure 2b) show good agreement results between the modeling and measurements for AWB dilbit products, but may slightly underestimate the droplets with nominal diameter of 390.7m. Experimental results of replicates 1 and 3 are close, but the results of replicate 2 show more droplets generated for nominal size classes 281 to 390.7m during the experiment. The predicted volume p.d.f. (probability density function) of CLB product (Figure 3a) showed good agreement with experimental data for droplet sizes from280.6m to461 m, especially with data from replicate 2, but slightly overestimated the function for nominal size classes smaller than 201.5m. Results in Figure 2 and 3 indicate that the modeling results represent the overall trend of the droplet size distribution resulting from surface spills of viscous oils, demonstrating that the VDROP model and our methods of modeling the droplet size distribution produced by waves are successful.

DISCUSSION:

The energy dissipation rate varies with time during wave breaking (Boufadel et al., 2008). While high energy rates tend to cause breakup, the duration of this high energy needs to be long enough to affect an actual breakup. Alternatively, a low energy rate cannot cause breakup no matter how long it is applied to the droplets. Therefore, one needs to consider the product of sufficiently high energy and duration to come up with meaningful design parameters. This

455

174 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE concept is similar to Chick’s law for disinfection where the product of disinfectant concentration and contact time with microorganisms is used for design (Metcalf and Eddy, 1972; Montgomery, 1985). For this reason, we elected to use a duration of 6s, and then to bring the energy dissipation rate down to zero for another 2 minutes (time between passes). The duration of 6 seconds represents approximately 3 wave periods, as previous studies have found that the mixing energy of wave breakers dissipates within 3 to 4 wave periods (Chen et al., 1999; Rapp and Melville, 1990).

The plots of cumulative volume fraction (Figure 2b and Figure 3b) may seem to indicate opposing biases for the two products, where the modeling results are smaller than the experiments for AWB (Figure 2b), and larger than the measurements for CLB (Figure 3b). However, by recognizing uncertainty in the experimental results (shown by variability among the replications) and by comparing with the plots of volume p.d.f. (Figure 2a and Figure 3a), one may find the model results are not biased. Taking Figure 3 as an example, the modeling results (Figure 3a) lie between the replicate data in the size range of 201.5 - 461 m, while the plot of all the modeling results larger than the experiment in Figure 3b is only caused by the overestimated amount of droplets of nominal size 201.5 m or smaller. Properties of dilbit products are clearly different from those of the conventional crude oils, which may have different dispersion behavior as mentioned before.

The maximum droplet diameter that the LISST particle-size analyzer can detect is 461 m; however, oil droplets larger than this size may exist in the water column of the wave tank experiments. Results presented in the previous section were generated using 461 microns as the maximum droplet size in the model simulation. To evaluate the effects of larger maximum droplet sizes (larger than 461 m) on the modeling results, two additional cases were simulated with maximum droplet diameters of 1 mm and 2 mm and these simulations were repeated to consider both dilbit products. Results from simulations for 1-mm and 2-mm maximum droplet diameters are almost identical, whether for the AWB (Figure 4a) or CLB (Figure 4b) product type.

For the AWB dilbit product (Figure 4a), comparing the supplemental simulation results with the case of initial droplet size of 461 m, oil droplets larger than 461 m are predicted when using a larger maximum droplet size. For the CLB product (Figure 4b) with lower oil viscosity, the size-distribution profiles from the supplemental simulations for 1-mm and 2-mm maximum droplet diameter are similar to that from the 461-m case, but also predicted is the presence of a small amount of oil droplets larger than 461 m. Comparing Figure 4a and 4b, the effect of the maximum droplet size was larger for the model of a higher viscosity oil. Further studies are needed to explore additional droplet-size measurement approaches and incorporation of experimental measurements of the droplet size distribution with model simulations of high viscosity dilbit products.

CONCLUSION:

A numerical droplet formation model, VDROP, is presented in this paper to simulate the droplet size distribution of diluted bitumen (dilbit) following a hypothetical oil spill in the marine environment. Two dilbit products (Access Western Blend and Cold Lake Blend) were studied in

456

175 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE the wave tank facility of the Department of Fisheries and Oceans (DFO) of Canada using experimental setups representative for a spill on the surface of ocean waters. Analogous to the experimental setup, the model was configured to simulate oil droplet size distribution with special consideration of droplet-breakup processes due to wave effects on surface oil spills. Modeling results showed good agreement with experimental data for both dilbit products with no dispersant presence. Further efforts are still required for including the dispersant effects on oil dispersive processes in simulations of oil spills in the open water. This study confirms the successful use of a droplet formation model for the simulation of droplet-size distribution resulting from effects of breaking waves on a surface oil spill.

REFERENCES:

Azbel, D.S., 1981. Two phase flows in chemical engineering. Cambridge University Press.

Bandara, U.C., Yapa, P.D., 2011. Bubble Sizes, Breakup, and Coalescence in Deepwater Gas/Oil Plumes. Journal of Hydraulic Engineering 137, 729-738.

Boufadel, M.C., Wickley-Olsen, E., King, T., Li, Z., Lee, K., Venosa, A.D., 2008. Theoretical foundation for predicting dispersion effectiveness due to waves, International Oil Spill Conference. American Petroleum Institute, pp. 509-513.

Calabrese, R.V., Chang, T., Dang, P., 1986. Drop breakup in turbulent stirred‐tank contactors. Part I: Effect of dispersed‐phase viscosity. AIChE Journal 32, 657-666.

Chen, G., Kharif, C., Zaleski, S., Li, J., 1999. Two-dimensional Navier-Stokes simulation of breaking waves. Physics of Fluids 11, 121-133.

Coulaloglou, C.A., Tavlarides, L.L., 1977. Description of interaction processes in agitated liquid- liquid dispersions. Chemical Engineering Science 32, 1289-1297.

Delvigne, G.A.L., Sweeney, C.E., 1988. Natural Dispersion of Oil. Oil & Chemical Pollution 4, 281-310.

Frisch, U., Parisi, G., 1985. Fully developed turbulence and intermittency. Turbulence and predictability in geophysical fluid dynamics and climate dynamics 88, 71-88.

Frisch, U., Sulem, P.-L., Nelkin, M., 1978. A simple dynamical model of intermittent fully developed turbulence. Journal of Fluid Mechanics 87, 719-736.

Gartner, J.W., Cheng, R.T., Wang, P.F., Richter, K., 2001. Laboratory and field evaluations of the LISST-100 instrument for suspended particle size determinations. Marine Geology 175, 199-219.

Hinze, J., 1955. Fundamentals of the hydrodynamic mechanism of splitting in dispersion processes. AIChE Journal 1, 289-295.

457

176 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE Kennard, E.H., 1938. Kinetic theory of gases. McGraw-hill New York. Kolmogorov, A.N., 1941. The local structure of turbulence in incompressible viscous fluid for very large Reynolds number. Dokl. Akad. Nauk. SSSR 30, 301-305. Reprinted in 1991: Proc. R. Sco. Lond. A1434, 1999-1913.

Li, Z., Lee, K., King, T., Boufadel, M.C., Venosa, A.D., 2008. Assessment of chemical dispersant effectiveness in a wave tank under regular non-breaking and breaking wave conditions. Marine pollution bulletin 56, 903-912.

Luo, H., Svendsen, H.F., 1996. Theoretical model for drop and bubble breakup in turbulent dispersions. AIChE Journal 42, 1225-1233.

Metcalf and Eddy, I., 1972. Wastewater Engineering: Collection, Treatment, Disposal. McGraw- Hill Inc., US.

Montgomery, J.M., 1985. Water Treatment: Principles and Design. John Wiley & Sons, New York.

Prince, M.J., Blanch, H.W., 1990. Bubble coalescence and break‐up in air‐sparged bubble columns. AIChE Journal 36, 1485-1499.

Rapp, R.J., Melville, W.K., 1990. Laboratory measurements of deep-water breaking waves. Philosophical Transactions of the Royal Society of London. Series A, Mathematical and Physical Sciences A331, 735-800.

Serra, T., Casamitjana, X., Colomer, J., Granata, T.C., 2002. Observations of the particle size distribution and concentration in a coastal system using an in situ laser analyzer. Marine Technology Society Journal 36 59-69.

Sterling, M.C., Bonner, J.S., Page, C.A., Fuller, C.B., Ernest, A.N.S., Autenrieth, R.L., 2004. Modeling crude oil droplet-sediment aggregationin nearshore waters. Environmental Science & Technology 38, 4627-4634.

Taylor, G., 1934. The formation of emulsions in definable fields of flow. Proceedings of the Royal Society of London. Series A 146, 501-523.

Tsouris, C., Tavlarides, L., 1994. Breakage and coalescence models for drops in turbulent dispersions. AIChE Journal 40, 395-406.

Venosa, A.D., Kaku, V.J., Boufadel, M.C., Lee, K., 2005. Measuring energy dissipation rates in a wave tank, Proceedings of 2005 International Oil Spill Conference. American Petroleum Institute, Washington, DC.

Walter, J.G., Blanch, H.W., 1986. Bubble break-up in gas - liquid bioreactors: Break-up in turbulent flows. The Chemical Engineering Journal 32, B7-B17.

458

177 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE Wang, C., Calabrese, R.V., 1986. Drop breakup in turbulent stirred‐tank contactors. Part II: Relative influence of viscosity and interfacial tension. AIChE journal 32, 667-676.

Wickley-Olsen, E., Boufadel, M.C., King, T., Li, Z., Lee, K., Venosa, A.D., 2007. Regular and Breaking Waves in Wave Tank for Dispersion Effectiveness Testing, Proceedings of the Arctic and Marine Oil Spill, Edmonton Alberta, Canada.

Yang, C., Wang, Z., Yang, Z., Hollebone, B., Brown, C.E., Landriault, M., Feildhouse, B., 2011. Chemical fingerprints of Alberta oil sands and related petroleum products. Environmental Forensics 12, 173-188.

Zhao, L., Torlapati, J., Boufadel, M.C., King, T., Robinson, B., Lee, K., 2014. VDROP: A numerical model for the simulation of droplet formation from oils of various viscosities. Chemical Engineering Journal Submitted (CEJ-D-14-00789).

Death of the parent droplet d0 d0 0 i  i10 ),(),(  Btdntdn B

Birth of d1 from breakup of d0 Complementary droplet d2   tdndgddB ),()(),( dt 3 3 3 B i10001 2 0  ddd 1 ),(  Btdn 2 Bi 2 B  vvBV 10 )( Update at time ti  ),(),(  Btdntdn 1 i i11 B d2

di-1 di di+1 di+2

Figure 1: Example of mass conservation for droplet breakup process

0.014 (a) 0.012

0.01 0.008 AWB-exp-replicate1 AWB-exp-replicate2 0.006 AWB-exp-replicate3 AWB-exp-average Volume Volume p.d.f. 0.004 Modeling results 0.002 0 461 103.9 122.6 144.7 170.8 201.5 237.8 280.6 331.1 390.7 Droplet diameter (m)

459

178 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE

1 0.9 (b) 0.8 0.7 0.6 AWB-exp-replicate1 0.5 AWB-exp-replicate2 AWB-exp-replicate3 0.4 AWB-exp-average 0.3 Modeling results 0.2 0.1

Cumulative Cumulative volume fraction 0 461 103.9 122.6 144.7 170.8 201.5 237.8 280.6 331.1 390.7 Droplet diameter (m) Figure 2: Experimental and modeling droplet size distribution at 1.2 m downstream of the oil release point for AWB dilbit: a) volume probability density function; b) cumulative volume fraction 0.009 0.008 (a) 0.007 0.006 CLB-exp-replicate1 CLB-exp-replicate2 0.005 CLB-exp-replicate3 0.004 CLB-exp-average 0.003 Modeling results Volume p.d.f. Volume 0.002 0.001 0 461 103.9 122.6 144.7 170.8 201.5 237.8 280.6 331.1 390.7 . Droplet diameter (m)

1 0.9 (b) 0.8

0.7 CLB-exp-replicate1 0.6 CLB-exp-replicate2 0.5 CLB-exp-replicate3 0.4 CLB-exp-average 0.3 Modeling results 0.2 0.1 0 461 103.9 122.6 144.7 170.8 201.5 237.8 280.6 331.1 390.7 Cumulative volume fractionvolume Cumulative Droplet diameter (m)

460

179 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE Figure 3: Experimental and modeling droplet size distribution at 1.2 m downstream of the oil release point for CLB dilbit: a) volume probability density function; b) cumulative volume fraction.

0.014 AWB - Initial droplet size - 461 micron 0.012 (a) AWB - Initial droplet size - 1 mm 0.01 0.008 AWB - Initial droplet size - 2 mm 0.006

Volume Volume p.d.f. 0.004 0.002 0 461 750 1000 1250 1500 1750 2000 103.9 122.6 144.7 170.8 201.5 237.8 280.6 331.1 390.7 Droplet diameter (m)

0.009 0.008 CLB - Initial droplet size - 461 micron (b) 0.007 CLB - Initial droplet size - 1 mm

0.006 CLB - Initial droplet size - 2 mm 0.005 0.004 0.003 Volume Volume p.d.f. 0.002 0.001 0 461 750 1000 1250 1500 1750 2000 103.9 122.6 144.7 170.8 201.5 237.8 Droplet280.6 diameter331.1 390.7 (m) Figure 4: Comparison of the predicted droplet size distribution using Vdrop model with the variation of initial droplet size. Table 1 Properties of weathered dilbit products Viscosity Density Surface Tension Oil (cp) (g/cm3) (mN/m) Access Western Blend (AWB) 350 0.955 30 Cold Lake Blend (CLB) 175 0.947 28

Table 2 Results of replicate experiments d d (average DSD) Absolute relative Oil Experiments 50 50 (μm) (μm) difference

461

180 300315 2014 INTERNATIONAL OIL SPILL CONFERENCE A B Abs((A-B)/B)

replicate1 416 0.5% AWB replicate 2 406 414 2.0% replicate 3 416 0.5% replicate 1 398 18% CLB replicate 2 330 336 1.8% replicate 3 312 7.1%

462

181