Year-round algal toxin exposure in free-ranging sea lions

Harmful algal bloom toxins cause illness and mortality in marine mammals worldwide, yet the potential for year-round exposure to these toxins has not previously been studied. We measured concentrations of domoic acid and saxitoxin in scats from Steller sea lions Eumetopias jubatus (n = 383 scats) and California sea lions Zalophus californianus (n = 125 scats) over a 2 yr period. Toxin concentrations in the scats were compared to the prey remains in the scats and to concentrations in nearshore bivalves. Saxitoxin was detected in 45% and domoic acid was detected in 17% of all scats tested, and both toxins were detected in all seasons and months of the year. A variety of benthic and pelagic fish were significantly associated with toxins in sea lion scats, including prey with low occurrence in the sea lions’ diet. Toxins detected in winter scats confirm that US West Coast marine mammals are exposed to domoic acid and saxitoxin through their prey outside of the expected algal bloom seasons.


MASTER'S THESIS
In presenting this thesis in partial fulfillment of the requirements for a master's degree at Western Washington University, I grant to Western Washington University the non-exclusive royalty-free right to archive, reproduce, distribute, and display the thesis in any and all forms, including electronic format, via any digital library mechanisms maintained by WWU.
I represent and warrant this is my original work, and does not infringe or violate any rights of others. I warrant that I have obtained written permissions from the owner of any third party copyrighted material included in these files.
I acknowledge that I retain ownership rights to the copyright of this work, including but not limited to the right to use all or part of this work in future works, such as articles or books.
Library users are granted permission for individual, research and non-commercial reproduction of this work for educational purposes only. Any further digital posting of this document requires specific permission from the author.
Pearson's residuals are shown for presence of prey item:level of toxin, with residuals >|1.96| being significant (in bold). Positive residual values indicate higher than expected level of toxin and negative residuals indicate lower than expected level of toxin given the presence of each prey item ..  Table 8. Significant Fisher's Exact tests (p<0.05) for 2 x 4 contingency, comparing presence/absence of prey to levels of no, low (<20 ng g -1 ), med (20-50 ng g -1 ), and high (>50 ng g -1 ) concentrations of toxin in the scat. Fisher's Exact p-values for the overall 2 x 4 contingency tables are presented. Post-hoc testing of all pairwise comparisons used a Bonferroni-adjusted alpha value to assess significance, where alpha was calculated as 0.05 divided by the number of pairwise comparisons. Significant 2 x 2 pairwise comparisons of presence of prey items at each toxin concentration level (e.g. presence of fish when low toxin levels in scat compared to presence of fish when no toxins in scat = Low:No) are presented by p-values with odds ratio and 95% confidence interval (CI

GENERAL INTRODUCTION
Harmful algal blooms (HABs) are increasing worldwide and have led to mortalities of both humans and marine wildlife, including several species of marine mammals (Fire & Van Dolah 2012). This trend is particularly concerning for declining and depleted wildlife populations (Forcada et al. 1999, Durbin et al. 2002, Shumway et al. 2003, Shearn-Bochsler et al. 2014, providing the need to understand what factors most influence exposure to these toxins. Despite recent mortality events related to HAB exposure and other pathogens, the US population of California sea lions Zalophus californianus is estimated at nearly 300,000 individuals (Carretta et al. 2014).
Conversely, the Steller sea lion Eumetopias jubatus experienced an 80% reduction in abundance in US waters from the 1970s through 1998 (Miller et al. 2005). The decline in the population was driven by declines in the western distinct population segment (DPS) (Western Alaska through Russia; west of 144° W) whereas during the same time period the eastern DPS (Southeast Alaska through California; east of 144° W) increased at 3% per year (Pitcher et al. 2007). Currently the western DPS is listed as endangered and the eastern DPS was recently delisted (NMFS 2008(NMFS , 2013. Though no mortalities related to algal toxins have been reported in Steller sea lions, monitoring to assess the threat of biotoxins is listed as a priority in the recovery plan of the western DPS (Goldstein et al. 2005, NMFS 2008) and for post-delisting monitoring of the eastern DPS (NMFS 2013).
This study assessed the year-round exposure of Steller and California sea lions in Washington State, USA, to two common algal toxins, domoic acid and saxitoxin, by measuring the concentrations of the two toxins in sea lion scat. To understand what 2 factors put marine mammals most at risk of marine algal toxin exposure, I review the history of HAB toxins in North America and their impacts on human health as well as the oceanographic conditions and life history features that lead to their occurrence. To more specifically understand the impacts of algal toxins on marine wildlife, I review our knowledge of illness and mortalities in marine mammals, potential routes of exposure through the marine food web, and sea lion food habits and foraging behavior.

HABs and Human Health
Toxins are produced by some species of bloom-forming marine algae in events referred to as harmful algal blooms (HABs) or red tides (Van Dolah 2000). Although not all HABs result in production of algal toxins, these toxins have perhaps become most well known for their contamination of shellfish and their impacts on human health (Van Dolah et al. 2001, James et al. 2010). On the US West Coast, the two most common outcomes of HABs that affect human and animal health are paralytic shellfish poisoning and amnesic shellfish poisoning (Horner et al. 1997, Trainer 2002. Paralytic shellfish poisoning is caused by dinoflagellates of the genera Alexandrium, Gymnodinium, and Pyrodinium (Backer & McGillicuddy 2006), though on the West Coast only species of Alexandrium have been documented (Trainer 2002).
These dinoflagellates produce saxitoxin and a suite of congeners, of which saxitoxin is the most toxic (Wang 2008, James et al. 2010. The primary producer of saxitoxin in Washington is Alexandrium catenella (Trainer 2002). Saxitoxin binds to site 1 of voltage-gated sodium channels embedded in the plasma membranes of nerve and muscle cells, effectively blocking neurotransmission by inhibiting sodium ion release and creation of an action potential, leading to paralysis (Bricelj & Shumway 1998, Wang 2008. In humans, saxitoxin initially causes numbness, particularly around the lips, mouth, and face, and in more severe cases can lead to a lack of motor coordination and respiratory paralysis (Backer & McGillicuddy 2006, Wang 2008. The lethal oral dose of saxitoxin in humans is 1 -4 mg; however, because saxitoxin is cleared through the urine in <24 hours, deaths are rare when medical treatment is given (Van Dolah et al. 2001, Backer & McGillicuddy 2006, James et al. 2010. Although the minimum dosage causing toxicity in humans is around 200 µg saxitoxin 100 g -1 tissue (2000 ng g -1 ), the regulatory limit for shellfish harvesting in the US is set conservatively at only 80 µg saxitoxin 100 g -1 tissue (800 ng g -1 ) (Wekell et al. 2004).
Amnesiac shellfish poisoning is primarily caused by diatoms of the genus Pseudo-nitzschia that produce the neurotoxic amino acid domoic acid (Landsberg 2002).
At least ten species of Pseudo-nitzschia have been identified along the Washington coast (Trainer et al. 2009), of which P. pseudodelicatissima, P. australis, P. cuspida, and P.
The toxin binds to glutamate receptors in neural tissue, primarily located in the hippocampus, causing continuous activation of the receptor and release of Ca 2+ ions until the cell dies, ultimately leading to atrophy of the hippocampus (Landsberg 2002. Clinical symptoms of amnesiac shellfish poisoning in humans include gastrointestinal distress, such as vomiting and diarrhea, and neurological 4 symptoms such as disorientation, seizures, and permanent memory loss (James et al. 2010, Fire & Van Dolah 2012. The minimum lethal oral dose estimated for an average 70 kg adult is 200 to 300 mg of domoic acid, or ingesting 0.7 -4 mg domoic acid kg -1 body weight (Wekell et al. 2004). Federal US regulators implemented a conservative regulatory limit of 20 µg g -1 (20,000 ng g -1 ) in the tissue of shellfish consumed by humans using the lowest observed dose to cause symptoms (50 mg), the assumption that a normal meal of shellfish is 200 g, and a safety factor of 12 (Wekell et al. 2004).
Outbreaks of both amnesiac and paralytic shellfish poisoning cause temporary and season-wide closures of shellfish harvest on Washington beaches (Trainer 2002, Dyson & Huppert 2010. The length of harvest closure depends in part on toxin depuration rates that vary among the species sampled from a period of days to weeks or months (Shumway 1990;Whyte et al. 1995). Blue mussels Mytilus edulis and California mussels Mytilus californianus are good indicators of saxitoxin in the environment because they display toxins weeks before other co-occurring bivalves (Bricelj & Shumway 1998).
Saxitoxin has few reported toxic effects on the mussels themselves, allowing them to accumulate the toxins quickly at high concentrations and to depurate the toxins from their tissues within several weeks (Shumway 1990, Bricelj & Shumway 1998. Similarly, domoic acid has no known toxic effects on mussels and they are able to completely clear the toxin from their body in two weeks or less (Novaczek et al. 1992, Whyte et al. 1995.
Conversely, Pacific razor clams Siliqua patula hold domoic acid in their tissues for as long as six to 18 months with relatively few toxic effects (Wekell et al. 1994, Trainer & Bill 2004. The length and timing of harvest closure and the impact of algal blooms on 5 human and wildlife health also depends on the oceanographic conditions and plankton life histories that result in bloom production.

Oceanography and Plankton Life History
HABs occur when the right oceanographic conditions and life history stages of the phytoplankton coincide (Van Dolah 2000). Relatively few species of phytoplankton produce toxins (Van Dolah 2000) and the ecological reasons for toxin production are not well understood (Turner & Tester 1997). Toxicity varies based on many factors including plankton species and strain, growth phase, salinity, temperature, nutrient limitation, and presence of bacterial flora (Anderson 1998, Van Dolah et al. 2003. Increased monitoring and surveillance of HABs has substantially increased our understanding of the species involved and their clinical effects (Landsberg 2002).
The apparent increase in HABs in recent years is in part attributed to increases in bloom-favorable conditions. Eutrophication and changes in climate and oceanographic conditions can enhance bloom growth and lengthen the bloom season , Moore et al. 2008. Furthermore, bloom-forming species may disperse to new areas via storms and currents (Anderson et al. 2002) and if climate change increases storm frequency or modifies currents it could increase dispersal of toxin-producing species (Moore et al. 2008). The unnatural dispersion of phytoplankton species via transport in 6 ship ballast water may also be increasing the dispersal of toxin-producing plankton (James et al. 2010).
The northwest Washington coast is a unique marine environment created by the confluence of oceanic currents with outflow from the Strait of Juan de Fuca, wind-and topographic-induced upwelling, and nutrient inputs from the Fraser and Columbia River outflows (Hickey & Banas 2003). A cold-water gyre, known as the Juan de Fuca Eddy, is formed seasonally off the mouth of the Strait of Juan de Fuca, spanning 50 km in diameter, and characterized by cold surface waters, low dissolved oxygen, and high nutrients (Hickey & Banas 2003, Marchetti et al. 2004, MacFadyen et al. 2005 High levels of domoic acid in nearshore shellfish along the Washington coast are thought to originate offshore in the Juan de Fuca Eddy, which has been identified as a hotspot for Pseudo-nitzschia and domoic acid production (MacFadyen et al. 2008, 7 Trainer et al. 2009). Although Pseudo-nitzschia cells have been found nearshore, the species assemblage differs from that found in the offshore eddy region and typically does not include the species known to produce toxin (Trainer et al. 2009). Waters from the eddy may be transported to the Washington coast as thin filaments that escape the southeastern margin of the eddy during southward, upwelling-favorable winds (MacFadyen et al. 2005, 2008, Trainer et al. 2009). During northward wind shifts, such as periods of low winds or storm events, eddy waters and associated Pseudo-nitzschia or domoic acid are brought onshore (MacFadyen et al. 2005, 2008, Trainer et al. 2002 Alexandrium species typically proliferate in calm, highly stratified waters, such as nearshore embayments, undergoing diel migrations to locate nutrients throughout the water column (Nishitani & Chew 1984, Anderson 1998, Trainer et al. 2003. Alexandrium blooms may occur with initial spring or summer warming, (in Washington the threshold is about 13˚ C, Nishitani & Chew 1984), and re-bloom in the fall when temperatures start to decline (Anderson 1998, Weise et al. 2002. Offshore occurrence of Alexandrium, including reports of deep-water cyst germination, has been reported in other areas and may in part be the result of vegetative cells or cysts transported from inshore to offshore waters via river outflow or coastal currents (Anderson 1998, Townsend et al. 2001. Marchetti et al. (2004) detected Alexandrium species in low concentrations in the Juan de Fuca Eddy, however, little is known about offshore occurrence of Alexandrium in the region.
Alexandrium are meroplankton, undergoing both sexual and asexual reproduction, including a vegetative, motile form and a dormant, resting cyst form (Wyatt & Jenkinson 1997, Anderson 1998. Gametogenesis leads to formation of zygotes that remain motile for 1-2 weeks before either producing new gametes or forming cysts, which settle to the 9 bottom and remain dormant until conditions for germination are right (Wyatt & Jenkinson 1997, Anderson 1998. Saxitoxin is produced by the bloom-forming vegetative cells, with the highest saxitoxin concentrations detected during the highest cell densities (Wyatt & Jenkinson 1997). However, dormant cysts may hold the toxin and are often more toxic than the vegetative cells (Wyatt & Jenkinson 1997, Bricelj & Shumway 1998). Even in relatively low abundance, Alexandrium may produce saxitoxin, as evidenced by detection of saxitoxin in nearshore shellfish despite Alexandrium cells not being detected in nearshore waters (Tweddle et al. 2010). While shellfish toxicity in the summer may be due to the vegetative cells, winter toxicity in shellfish is primarily attributed to cysts (Wyatt & Jenkinson 1997, Bricelj & Shumway 1998. Alexandrium cysts can be upwelled into the water column and taken up by nearshore shellfish (Schwinghamer et al. 1994, Tweddle et al. 2010, Butman et al. 2014. When these algal blooms or cyst hotspots overlap with the prey and foraging areas of marine mammals, there is the potential for toxin exposure and serious illness or mortalities.

HABs and Marine Mammals
Harmful algal bloom (HAB) toxins have been suspected or implicated in mass strandings of multiple species of marine mammals including dolphins, seals, sea lions, sea otters Enhydra lutris, West Indian manatees Trichechus manatus, and large whales (Bossart 2011, Fire & Van Dolah 2012. HAB-related strandings of marine mammals were first documented in the US in the 1980s (Anderson & White 1989) and have since become increasingly frequent (Fire & Van Dolah 2012). Today, the majority of marine mammal mass stranding events in the US have been attributed to HAB toxins (Fire et al. 2010). The specific concentrations that cause toxicity and mortality are largely unknown, but the unique biology of marine mammals may make them more susceptible to the toxins than humans or terrestrial wildlife (Anderson & White 1989).
Saxitoxin binds to marine mammal neural tissue with high affinity and specificity (Trainer & Baden 1999). The paralytic nature of saxitoxin may put marine mammals at particular risk of toxicity. As part of the mammalian dive response, marine mammals shunt blood to their vital tissues such as the lungs and brain and away from detoxifying tissues such as liver and kidney (Anderson & White 1989). Sublethal effects, such as loss of equilibrium or respiratory distress typical of paralytic shellfish poisoning in humans, could quickly incapacitate marine mammals and lead to death (Anderson & White 1989).
Although symptoms of chronic exposure have not been documented, Durbin et al. (2002) suggest that repeated exposure to saxitoxin could affect diving capabilities. The authors suggest that longer surfacing time after a dive, as observed in North Atlantic right whales Eubalaena glacialis, could be the result of saxitoxin exposure and inhibition of respiratory or muscle coordination and could increase the possibility of ship strike (Durbin et al. 2002).
Saxitoxin has caused mass mortalities of humpback whales Megaptera novaeangliae and endangered Mediterranean monks seals Monachus monachus (Anderson & White 1989). In 1987, 14 humpback whales died in Cape Cod due to eating Atlantic mackerel Scomber scombrus containing high concentrations of saxitoxin (Geraci et al. 1989). The whales appeared in good body condition and were observed actively 11 feeding up to 90 minutes before death (Geraci et al. 1989). In 1997, saxitoxin was suspected in the deaths of 117 Mediterranean monk seals off the coast of Western Sahara (Hernández et al. 1998). Similar to the case of the humpback whales, the seals appeared in good body condition, but exhibited a lack of motor coordination and paralysis, and many were found floating with congested lungs indicative of drowning (Hernández et al. 1998). In addition to these mortality events, saxitoxin exposure has also been documented in a variety of other pinniped and cetacean species (Durbin et al. 2002, Doucette et al. 2006, Lefebvre et al. 2016).
In general, little is known about the concentrations of saxitoxin that may cause health effects in marine mammals, and exposure has primarily been documented through testing feces, urine, and body fluids or testing prey species or stomach contents (Geraci et al. 1989, Durbin et al. 2002, Doucette et al. 2006. For terrestrial mammals, the lethal oral dose of saxitoxin ranges from 200 to 600 µg 100 g -1 (6,000 ng g -1 ) (Anderson & White 1989). In the case of the stranded humpback whales, it was estimated that the concentration of saxitoxin in the whale's prey was high enough that whales would obtain the lethal dose of saxitoxin before consuming their daily average amount of food (Anderson & White 1989). Although the majority of saxitoxin is cleared rapidly through the urine (Van Dolah et al. 2001), saxitoxin has been detected in low concentrations in body tissues such as liver, kidney, muscle and brain in Mediterranean monk seals (Hernández et al. 1998, Reyero et al. 1999. Because feces was not collected from these monk seals, it unknown how concentrations in the scat compared to accumulation in the body tissues. Cases of marine mammal strandings due to domoic acid have been much more extensive than those associated with saxitoxin. Mortalities have included California sea lions (Gulland 2000, Scholin et al. 2000), long-beaked common dolphins Delphinus capensis (Torres de la Riva et al. 2009), a common minke whale Balaenoptera acutorostrata (Fire et al. 2010), and sea otters (Torres de la Riva et al. 2009). Exposure to domoic acid has confirmed in a wide variety of other species (Fire et al. 2009, Lefebvre et al. 2010, Twiner et al. 2011, McHuron et al. 2013. Both acute and chronic pathologies of domoic acid toxicity have been described in marine mammals ). Acute symptoms include head weaving, muscle tremors, seizuring, and ataxia . Chronic symptoms are epilepsy with intermittent seizures and abnormal behaviors such as stranding in unnatural locations (e.g. roadways), obsessive repetitive behaviors, or abnormal aggression . Both acute and chronic pathologies may be detected through atrophy of neurons and distinct lesions in the hippocampus , Buckmaster et al. 2014. Chronic toxicity may compromise foraging behavior and navigation, causing sea lions to decrease diving ability and travel out of their normal range (Thomas et al. 2010). Domoic acid is also linked to reproductive failure, causing abortion, premature parturition, and stillbirth in California sea lions (Brodie et al. 2006).
Due to its water-soluble nature, domoic acid is primarily detected in feces, urine, and other fluids and has not been identified in other body tissues such as kidney, liver, muscle, and brain (Gulland 2000). Domoic acid is rapidly excreted from the body through both urine and feces and is thus commonly highest in these products (McHuron 13 et al. 2013). While concentrations in feces indicate exposure, detection of domoic acid in the urine or serum indicates absorption through the digestive tract and into the blood stream (Fire et al. 2010, Lefebvre et al. 2010. Still, it is not well known what concentrations induce toxicity. In sea lions exhibiting clinical symptoms of acute domoic acid toxicity, domoic acid was not always detected in feces, urine, or other fluids (Gulland 2000. The toxin concentrations detected in feces of sea lions exhibiting acute symptoms may be as low as 10 ng ml -1 ) to >120,000 ng g -1 (Lefebvre et al. 1999).

HABs and the Marine Food Web
Given the breadth of species with known exposure to HAB toxins and the wide variety of feeding habits of these species, it is likely that algal toxins transfer up the marine food web by many pathways. Domoic acid and saxitoxin have been detected in a variety of fish species , Vigilant & Silver 2007, suggesting that generalist foragers such as sea lions may be exposed to toxins through multiple prey items. The mechanism of marine mammal dietary exposure to algal toxins has been primarily investigated through sampling of prey or fecal remains of actively foraging or live-captured animals (Durbin et al. 2002) and sampling the feces or stomach contents of ill or deceased animals (Geraci et al. 1989, Lefebvre et al. 1999, Fire et al. 2010).
14 Several mass strandings of marine mammals due to HAB toxins have been linked to consumption of a specific prey species, while other studies have found more complex spatial and temporal trends. Stranded California sea lions and a common minke whale had eaten Northern anchovy Engraulis mordax contaminated with high concentrations of domoic acid (Lefebvre et al. 1999, Fire et al. 2010. In a mass stranding of humpback whales due to saxitoxin toxicity, the whales' normal prey item at that time of year, sand lance Ammodytes spp., was absent, leading the whales to travel farther north looking for prey and resulting in consumption of contaminated Atlantic mackerel (Anderson & White 1989). Spatial overlap of foraging areas and algal blooms is of particular concern for adult female California sea lions, which strand at a greater frequency from domoic acid toxicity compared to other demographic groups due to foraging in offshore locations that overlap with the major algal blooms (Silvagni et al. 2005. Several pathways for trophic exposure in the marine food web exist. Zooplankton and planktivorous fish are a major source of algal toxins to marine mammals and seabirds . Zooplankton, such as krill and copepods, have exposed humpback and blue whales Balaenoptera musculus to domoic acid  and North Atlantic right whales to saxitoxin (Durbin et al. 2002). Domoic acid and saxitoxin have been detected in anchovy, sardine Sardinops sagax, and Pacific herring Clupea pallasii (Lefebvre et al. 2001, Costa & Garrido 2004, Jester et al. 2009).
Saxitoxin in Pacific sand lance Ammodytes hexapterus has led to deaths of several marine birds (Shumway et al. 2003, Shearn-Bochsler et al. 2014. Toxins primarily accumulate in the gut or viscera of fish, with little accumulation in the muscle, and do not appear to cause direct toxic effects on the fish (Lefebvre et al. 2001, 2007, Jester et al. 2009).
Domoic acid and saxitoxin have also been detected in a variety of both benthic and pelagic fish species , Vigilant & Silver 2007. Vigilant et al. (2007) propose two possible pathways for exposure of benthic fish to these toxins. Bentho-pelagic feeding fish may eat planktivorous fish or zooplankton that are directly feeding on the toxic phytoplankton (Vigilant & Silver 2007). The second pathway for benthic feeding fish is feeding on plankton or fish fecal pellets containing the toxin when they sink to the bottom (Vigilant & Silver 2007). HABs toxins are also found higher in the marine food web in fish such as albacore tuna Thunnus alalunga .
Toxicological risk in marine mammals is frequently estimated using the known concentrations of toxins in fish tissues and calculating toxin exposure based on feeding rates (Fire et al. 2010 and bioenergetics models (Bejarano et al. 2007). In a somewhat novel approach, Jensen et al. (2015) used the otoliths found in harbor seal Phoca vitulina feces to selectively sample fish species to test for domoic acid and saxitoxin and estimate oral dosages from eating these common prey species. Bejarano et al. (2007) used a bioenergetics model including toxin concentrations measured in Northern anchovy and Pacific sardine to estimate the relative impact and toxicity of these two species on California sea lions. The authors concluded that the risk of domoic acid exposure is higher if sea lions ingest anchovy compared to sardine (Bejarano et al. 2007). However, both prey availability and the migration and foraging patterns of the predator itself will further influence the risk of exposure to algal toxins.

Sea Lion Food Habits and Foraging Behavior
Steller and California sea lions are considered generalist predators that feed on seasonally and locally abundant prey and that may migrate in response to seasonally predictable dense prey aggregations such as herring spawns and salmon returns (Weise 2000, Sinclair & Zeppelin 2002, Womble et al. 2005, Bredesen et al. 2006. A more diverse diet could reduce the potential for acute HAB toxicity from ingesting a single contaminated prey species, however the detection of HAB toxin in a variety of pelagic and benthic fish , Vigilant & Silver 2007 suggests that generalist predators may be at risk of exposure from multiple vectors. Further, although sea lions may feed on a large number of prey species, diets are often dominated by only a few species , Riemer et al. 2011).
The overlapping range of eastern DPS Steller sea lions and California sea lions leads to an overlap in diet, suggesting the two species could have a similar risk of exposure to algal toxins. Steller sea lions of all demographic groups are present on the Washington coast year-round and in recent years classic breeding structure and low levels of pupping have been observed (Jeffries et al. 2000, NMFS 2013, Wiles 2015. California sea lions breed in southern California and the Gulf of California, Mexico, and males disperse north as far as Alaska in the non-breeding season (Maniscalco et al. 2004).
California sea lions occur in Washington year-round, with peak abundance in the nonbreeding season (Antonelis & Fiscus 1980, Jeffries et al. 2000, Edgell & Demarchi 2012. Typically only males are seen in Washington waters (Jeffries et al. 2000), with few known cases of adult females (Thomas et al. 2010).
Diet in sea lions differs by age (Orr et al. 2011), sex (Trites & Calkins 2008), season, year, and location (Sinclair & Zeppelin 2002, McKenzie & Wynne 2008, Riemer et al. 2011 and is influenced by abundance and distribution of prey ). Steller sea lion food habits vary greatly by study region, but top prey species include walleye pollock Theragra chalcogramma and Atka mackerel Pleurogrammus monopterygius in Alaska , Sinclair & Zeppelin 2002 and Pacific hake Merluccius productus in California, Oregon, and Washington (Scordino 2010, Riemer et al. 2011  . Differences in diet also stem from differences in foraging area by age, sex, and season (Merrick et al. 1994, Merrick & Loughlin 1997. Adult female Steller sea lions forage over the continental shelf, closer to shore with shorter trip durations in the summer (17 km) compared to winter (133 km) (Merrick & Loughlin 1997, Trites & Calkins 2008, whereas the foraging behavior of adult male Steller sea lions has not been studied (Trites & Calkins 2008). This shift in female foraging distance is in part attributed to females with dependent pups in the summer limiting forage trip distance and due to shifts in prey distribution in the winter (Merrick & Loughlin 1997). Adult male California sea lions typically forage over the continental shelf, but in a warm water El Niño event, sea lions foraged longer distances (averaged 124 km compared to 48 km in pre-El Niño years) and further offshore (Weise et al. 2006). near the shelf break, with sea lions occurring closer to shore in the summer. Given the overlap between the occurrence of HAB blooms on the Washington coast and the range of these two sea lion species, there is a potential for exposure to HAB toxins through the marine food web.

Research Goals and Purpose
In this study, I used fecal analysis to assess the year-round exposure of sea lions in Washington State to domoic acid and saxitoxin by measuring the concentrations of the two toxins in scat samples. There were three research goals in this study. The first was to determine to what degree sea lions in Washington are exposed to HABs and to assess the temporal and spatial factors affecting the occurrence (presence/absence) of HAB toxins in scat. The second goal was to identify potential routes of trophic exposure to algal toxins by looking for differences in the food habits of sea lions with and without detectable toxins in the scat and to investigate whether particular prey are more often associated with the presence of toxin than others. Lastly, I attempted to compare the presence and concentrations of toxins in sea lion scat to those measured in nearshore bivalves over the same time period to determine whether bivalves could be used to predict HAB exposure in sea lions. To

INTRODUCTION
Marine wildlife is frequently exposed to a variety of pollutants through ingestion of contaminated prey. These pollutants may result from human activities, such as organochlorine pesticides and industrial organic chemicals (Tanabe 2002, Law et al. 2003, Islam & Tanaka 2004 or from elevated levels of essential and non-essential elements (Sydeman & Jarman 1998, Law et al. 2003, Scheuhammer et al. 2007).
Pollutants may also be naturally occurring, such as toxins produced by bloom-forming marine algae (Van Dolah 2000). Harmful algal bloom (HAB) toxins have led to mass mortalities of fish and seabirds (Work et al. 1993, Sierra Beltrán et al. 1997, Van Dolah 2000, Shumway et al. 2003 and are suspected or implicated in mass strandings of many species of marine mammals (Geraci et al. 1989, Lefebvre et al. 1999, 2010, Reyero et al. 1999, Gulland 2000, Kreuder et al. 2003, Torres de la Riva et al. 2009, Fire et al. 2015. A number of studies have examined algal toxin exposure in live, free-ranging marine mammals including several species of cetaceans (Durbin et al. 2002, Doucette et al. 2006) and pinnipeds , Lefebvre et al. 2016; however, none have directly examined the potential for year-round exposure through the food web.
Two phytoplankton taxa implicated in marine mammal mortalities are diatoms of the genus Pseudo-nitzschia and dinoflagellates of the genus Alexandrium, which produce 22 domoic acid and saxitoxin, respectively (Van Dolah et al. 2001). In humans, domoic acid causes amnesiac shellfish poisoning (ASP), while saxitoxin, and its many congeners, cause paralytic shellfish poisoning (PSP) (James et al. 2010). On the US West Coast, algae blooms leading to unsafe levels of domoic acid and saxitoxin cause annual closures of shellfish harvests (Horner et al. 1997, Trainer 2002, Trainer et al. 2007 were recently documented in a variety of marine mammals in Alaska, including some cases of concurrent exposure having both toxins detected (Lefebvre et al. 2016).
HABs appear to be increasing in frequency and severity worldwide (Van Dolah 2000) giving rise to the need to understand what factors most influence exposure to HAB-related toxins in marine wildlife, particularly in declining populations (Forcada et al. 1999, Durbin et al. 2002, Shumway et al. 2003, Shearn-Bochsler et al. 2014). The Washington coast is a unique marine environment for studying domoic acid and saxitoxin, which both impact state shellfish fisheries (Trainer et al. 2002(Trainer et al. , 2003. A hotspot for Pseudo-nitzschia blooms on the Washington coast is the Juan de Fuca Eddy, a seasonally formed, cold-water gyre located offshore the mouth of the Strait of Juan de Fuca (Trainer et al. 2002). The eddy forms in the spring with southward upwellingfavorable winds and dissipates in the fall (MacFadyen et al. 2005(MacFadyen et al. , 2008. Some waters escape from the southeastern margin of the eddy bringing their associated toxins to the 23 nearshore environment (MacFadyen et al. 2005, 2008, Trainer et al. 2009). In contrast, Alexandrium blooms occur nearshore, typically in embayments with highly stratified waters, after water temperatures begin warming in the spring (Nishitani & Chew 1984, Anderson 1998, Trainer et al. 2003). These differences in bloom dynamics may manifest as different food web pathways to top predators, such as sea lions, depending on the specific prey vectors, predator foraging area, and seasonal and annual changes in both.
Despite mortality events related to HAB exposure and other pathogens, the US population of California sea lions is estimated at nearly 300,000 individuals (Carretta et al. 2014). Conversely, the Steller sea lion Eumetopias jubatus, experienced an 80% reduction in abundance from the 1970s through 1998 in US waters (Miller et al. 2005).
The decline in the population was driven by declines in the western distinct population Illness and mortality of marine mammals due to HAB toxins are frequently linked to consumption of specific prey species, but some studies have found more complex 24 spatial and temporal trends. Northern anchovy Engraulis mordax was named the primary vector of toxin exposure in the deaths of California sea lions and a common minke whale Balaenoptera acutorostrata that died due to domoic acid toxicity (Lefebvre et al. 1999, Scholin et al. 2000, Fire et al. 2010. In a mass stranding of humpback whales Megaptera novaeangliae due to saxitoxin toxicity, the whales' normal prey item at that time of year, sand lance Ammodytes spp., was not present, leading the whales to travel farther north looking for prey and resulting in consumption of contaminated Atlantic mackerel Scomber scombrus (Anderson & White 1989, Geraci et al. 1989. Spatial overlap of foraging areas and algal blooms is of particular concern for adult female California sea lions, which strand at a greater frequency due to domoic acid toxicity compared to other demographic groups (Silvagni et al. 2005. Adult male California sea lions forage close to rookeries during the breeding season and then migrate north during the non-breeding season, but adult females remain in California waters throughout the year and forage in offshore locations that overlap with the major algal blooms (Silvagni et al. 2005). Thus, a combination of prey availability and the migration and foraging patterns of the predator itself influence the risk of exposure to algal toxins.
Sea lions are considered generalist predators that change their diet based on seasonally and locally abundant prey and may migrate in response to seasonal prey fluctuations (Weise 2000, Sinclair & Zeppelin 2002, Womble et al. 2005, Bredesen et al. 2006). In other areas the diet of Steller sea lions , Sinclair & Zeppelin 2002, Bredesen et al. 2006, Riemer et al. 2011) and California sea lions (Lowry et al. 1991, Weise & Harvey 1999, Orr et al. 2011 is often dominated by only a few species, which is not the case on the Washington coast where diet specialization in both Steller and California sea lions is low ). Sea lion diets are diverse, with as many as 76 prey identified (McKenzie & Wynne 2008) and as many as 25 different prey found in a single scat (Riemer et al. 2011). Domoic acid and saxitoxin have been detected in both benthic and pelagic fish species , Vigilant & Silver 2007, suggesting that while a more diverse diet could lessen the effects of acute toxicity experienced by ingesting a single infected prey species, generalist predators are at risk of exposure from multiple prey items. Previous studies have attempted to examine the question of marine mammal exposure to algal toxins through sampling of prey and fecal remains of actively foraging or live-captured animals (Durbin et al. 2002, Doucette et al. 2006, through sampling of plankton and nearshore bivalves, such as the blue mussel Mytilus edulis, in areas near mortality events as indicators of blooms and toxin production (Scholin et al. 2000, Bargu et al. 2010, and through sampling the feces or stomach contents of ill or deceased animals (Geraci et al. 1989, Lefebvre et al. 1999, Fire et al. 2010. In this study, I assessed the year-round exposure of free-ranging sea lions in Washington State to domoic acid and saxitoxin through comparing the toxin concentrations measured in feces to the prey remains in the scat and to toxin concentrations measured in nearshore bivalves collected over the same time period. The goals of this study were to 1) determine year-round baseline levels of HAB toxins ingested by both sea lion species and to identify factors associated with toxin exposure, including season, year, and haulout location 2) determine if there was a difference in diet 26 between sea lions with and without measureable levels of toxins and whether particular prey were more associated with toxin exposure, and 3) compare the presence and concentration of toxins in the scat to that in nearshore bivalves to determine how well these nearshore indicators predicted toxin exposure of top predators. To the best of my knowledge, this is the first study to document domoic acid and saxitoxin in free-ranging sea lions on the US West Coast.

Study Area
The study area was located at the northernmost extent of the California Current on the northwest coast of Washington State (MacFadyen et al. 2008). The region is characterized by high phytoplankton biomass and primary productivity, driven by wind and topographic induced upwelling, as well as the seasonally formed Juan de Fuca Eddy, a cold water gyre located off the mouth of the Strait of Juan de Fuca, and nutrient inputs from the Fraser and Columbia River outflows (Horner et al. 1997, Trainer et al. 2002, Hickey & Banas 2003

Toxin Analysis
After collection, scats were processed fresh or were frozen and later thawed for processing. I subsampled the scats for toxin analysis by placing approximately 4 g of each scat in a 15 ml centrifuge tube. To ensure that no prey remains were lost during subsampling, the 4 g of fresh scat was pushed through either 1) a nylon, fine mesh, paint strainer bag (0.25 mm), or 2) through a 0.5 mm sieve to be collected for toxin analysis. If the scats were frozen prior to subsampling, I collected as much of the liquid as possible from the scat as it was pushed through the mesh in an attempt to capture the watersoluble toxins. I weighed, labeled, and placed each centrifuge tube in a labeled Whirl-Pak bag then frozen at -20˚ C. At least 2 g of feces were necessary for the analysis of each toxin. When scats were small and had insufficient material for analyzing both toxins, I prioritized the analysis of domoic acid because it has caused strandings and mortalities of sea lions on the US west coast (Scholin et al. 2000) and there is particular interest in determining the exposure of Washington sea lions to this toxin.
Scat subsamples were sent to Dr. Elizabeth Frame at the Wildlife Algal Toxin Research and Response Network at NOAA's Northwest Fisheries Science Center in Seattle, WA, where they were analyzed using enzyme-linked immunosorbent assays (ELISA). The extraction methods for the toxin assays followed methods described in Lefebvre et al. (2016). Scat samples were thawed and split into two 15-ml falcon tubes for domoic acid and saxitoxin analysis respectively, with approximately 2 g in each tube.
For domoic acid analysis, the sample was extracted using 50 % methanol and 50 % water. For saxitoxin, the extraction solvent was 80 % ethanol and 20 % water. Samples were homogenized and centrifuged at 4˚C and the resulting supernatant was filtered and spun in a microcentrifuge. Domoic acid was analyzed using ASP direct cELISA kits made by Biosense Laboratories (Norway). Saxitoxin was analyzed using Saxitoxin (PSP) ELISA kits made by Abraxis Laboratories (Pennsylvania). Samples were diluted using a 10 % methanol buffer at a 1:100 dilution for domoic acid analysis and a 1:50 dilution for saxitoxin analysis. The lower quantification limits for the ELISA's are 3 ng g -1 for saxitoxin and 4 ng g -1 for domoic acid.
The highest concentration of domoic acid measured in this study (672.2 ng g -1 ) was detected in a California sea lion scat. Due to an error in sample processing, this same concentration was reported in two consecutive scats that appeared to be a single sample erroneously processed twice for domoic acid and recorded with two different identifications. Because I was unable to identify the sample, it was discarded from statistical analyses and diet comparisons.

Food Habits Analysis
After subsampling for the toxin analysis, the remaining fecal material was processed for food habits analysis. For the majority of samples, I followed the methods of Orr et al. (2003), using a commercial washing machine to clean the fresh fecal material from the prey remains (cephalopod beaks and bones and other hard parts from fish).
Fresh or thawed scats were washed from the Whirl-Pak™ bags into nylon, fine mesh (<0.05 mm) paint strainer bags. The paint strainer bags were washed in a washing machine with cold water, on a regular cycle, using a small amount of detergent to help break down the oily fecal material. Following washing, the prey remains were emptied from the mesh bags into a 0.5-mm sieve and hard parts were picked out using forceps.
Samples with gravel were washed by hand through nested sieves of 2 mm, 1mm, and 0.5 mm, following methods described by Lance et al. (2001), so that the gravel would not damage prey remains during cleaning (Orr et al. 2003 (Lance et al. 2012), therefore higher taxonomic classifications were used in these cases.
Salmon Oncorhynchus spp. were identified during food habits analysis by size and grouped as "juvenile salmon" (smolts) and "non-juvenile salmon" (all other age classes) by comparing to references of different species and age classes. Remains that could not be identified to a taxonomic group were recorded as unidentified bony fish (class Osteichthyes). Prey taxa recorded as present/absent for each sample were converted to percent frequency of occurrence (FO) in sea lion diet, where FO of a particular taxon is equal to the number of scats having that taxon divided by the total number of scats with any identifiable prey.

Bivalve Samples
Biotoxin results measured in nearshore bivalves from January 2011 through  Parentheses refer to sites having more than one sample location.

Analysis of shellfish was performed by the WDOH Public Health Laboratory in
Seattle, WA, using the standardized mouse bioassay for detection of saxitoxin (AOAC 1965, APHA 1980, NSSP 2013 and high-performance liquid chromatography for detection of domoic acid (Quilliam et al. 1995). Saxitoxin is reported in µg saxitoxin equivalents per 100 g of tissue with a regulatory limit for human consumption of 80 ug 100 g -1 , or 800 ng g -1 (Wekell et al. 1994). Domoic acid is reported in µg domoic acid g -1 of tissue with a regulatory limit of 20 µg g -1 , or 20,000 ng g -1 (Wekell et al. 1994).
Samples reported by WDOH as not detected for both toxins are reported here as "0". For saxitoxin, concentrations below 38 µg 100 g -1 are reported by WDOH as "<38" and are reported here conservatively as 380 ng g -1 . This value was chosen because no true concentrations were measured at 380 and because I do not know where between zero and 380 the true concentration lies. For domoic acid, WDOH reports concentrations of both <1 µg g -1 and 1 µg g -1 (1000 ng g -1 ), hence I substituted <1 µg g -1 for 0.90 µg g -1 , or 900 ng g -1 . This number was chosen to differentiate between a true concentration of 1000 and <1000 and because I do not know where between zero and 1000 the true concentration lies. Only two species of shellfish are included in this analysis: California mussel Mytilus californianus and Pacific razor clam Siliqua patula. California mussels were the dominant species collected at all sites except Kalaloch Beach (Figure 2) where razor clams were dominant.
Nearshore bivalve toxin content was graphically compared to the concentrations in sea lion scats to examine the relationship between presence and concentration of toxin in bivalves to that in the scats. Bivalves harvested for human consumption are sampled on a weekly or bi-weekly time scale based on blooms that peak and decline in short time periods (Horner and Postel 1993;Trainer 2002) and toxin depuration rates of days to weeks in the species monitored (Shumway 1990;Whyte et al. 1995). California mussels depurate saxitoxin in a span of three weeks to >12 weeks (Bricelj and Shumway 1998), whereas they depurate domoic acid in as little as two weeks (Whyte et al. 1995).
Although high concentrations of saxitoxin have been detected in razor clams (Bricelj and Shumway 1998;Trainer 2002), razor clams typically accumulate only low levels of saxitoxin (Anderson et al. 2002, Trainer et al. 2003 and depuration rates for the toxin have not been reported. Razor clams hold domoic acid in their tissues for as long as 6-18 months (Wekell et al. 1994); therefore, the detection of domoic acid in razor clams over or class). Salmon were grouped as "juvenile" and "non-juvenile" salmon.
For the two sea lion species, I ran NMDS and ANOSIM for each toxin separately.
I used the vegdist function with binary Jaccard distance matrix, employed to compare presence/absence of species (Riemer et al. 2011), to calculate the dissimilarity matrix.
For NMDS, I used the metaMDS function, which can handle zero distance where two points (two scats) are identical (Oksanen et al. 2015). Each scat represents a single point in the NMDS and each individual prey item was input as present/absent. The plotted points were labeled first by the presence/absence (Yes/No) of toxin measured in the scat, then in a second MDS by categorical levels, "no", "low", "med", and "high" toxin.
Designation of toxin levels was based on natural breaks in the concentrations measured in scats and not intended to be indicative of toxicological effects. For both toxins, "no" represents all samples below detection limit, "low" between 0-20 ng g -1 , "med" between >20-50 ng g -1 , and "high" for concentrations >50 ng g -1 . MDS ordination was evaluated by stress values as defined by Kruskal (1964). I tested for significant differences in diet based on presence/absence of toxin using ANOSIM, where p < 0.05 indicated a significant difference.
To determine the relationship between specific prey items and toxin exposure in sea lions, I used 2 x 2 and 2 x 4 contingency tables to compare the presence/absence of individual prey items to either presence/absence of toxin (2 x 2) or categorical toxin levels as defined above (2 x

Toxin Detection
A total of 508 scats were collected; 383 scats were from Steller sea lions Eumetopias jubatus and 125 scats were from California sea lions Zalophus californianus (Table 1)

Toxin Prevalence by Season, Year, and Location
Logistic regression analysis identified several explanatory models for the presence of toxins in Steller sea lions, but for California sea lions the regression models were not successful. In the case of California sea lions, the intercept model had the lowest AIC among all models tested (season, year, and season+year) for both domoic acid and saxitoxin.
In Steller sea lions, predicting presence of saxitoxin by season and year, the most parsimonious model was the full model including season, year, and their interaction (Table 2). Saxitoxin exposure in year 1 was higher in spring, fall, and winter compared to year 2, while year 2 had greatest saxitoxin exposure in the summer ( Figure 6A).
Considering domoic acid, the best model included the additive effect of season and year (Table 3). Domoic acid exposure peaked in the summer of both years, but appeared higher in the spring and summer of year 2 compared to year 1 ( Figure 6C).

Considering haulout complex (Tatoosh and Carroll/Sea Lion Rock), season
(spring and fall), and year, the most parsimonious model for predicting saxitoxin was the full model including complex, year, and their interaction (Table 3). In year 1, saxitoxin exposure was higher at the Tatoosh Island complex compared to Carroll/Sea Lion Rock, while in year 2 the exposure rates appeared similar ( Figure 6B). For domoic acid, the best model included haulout complex only (Table 3). Domoic acid exposure was higher at the Carroll/Sea Lion Rock complex compared to the Tatoosh Island complex ( Figure 6D).

Sea Lion Diet
A total of 39 prey taxa (lowest taxonomic group) was identified in Steller sea lion scats (Table 4) and 30 prey taxa were identified in California sea lion scats (Table 5).
Eight scats from Steller sea lions contained no identifiable prey remains (unidentified bony fish) and were removed from diet analyses. The most common prey species (>20 %  (Table 4). Flatfishes were almost completely absent in winter samples (1% FO).
For California sea lions, due to small sample sizes in spring (n = 20, year 1 only) and summer (n = 19, year 2 only), I compared diet between years during fall season only.
Most noticeably, hake consumption dropped dramatically from fall of year 1 to year 2, reduced from FO of 53% to only 8% (Table 5). For Steller sea lions, I compared diet between the haulout complexes with the greatest sample size, Tatoosh Island (n = 97) and Carroll/Sea Lion Rock (n = 100), for the spring and fall seasons when both sites were sampled. Clupeids and salmonids were more common in samples from Tatoosh Island (67% and 57%, respectively) compared to Carroll/Sea Lion Rock (45% and 22%). Skates and flatfishes were higher at Carroll/Sea Lion Rock (56% and 40%) compared to Tatoosh (40% and 7%). Codfishes were more common at Tatoosh (29%) compared to Carroll/Sea Lion Rock (12%), primarily driven by the high occurrence of walleye pollock in the spring of year 1 (Table 4).

Diet Influences on Toxin Prevalence
There were no convergent solutions using nonmetric multidimensional scaling (NMDS) to examine the diet of either Steller or California sea lions (Figures 7 and 8). I present the lowest stress values calculated after 20 NMDS iterations, which between 0.1-0.2 would typically represent good ordination (Kruskal 1964), however due to nonconvergence do not represent a true overall minimum stress or fit to he data. Analysis of similarity (ANOSIM) found a significant difference in Steller sea lion diet between scats with and without saxitoxin (R = 0.03, p = 0.001) and between scats with toxins grouped by concentration level (saxitoxin: R = 0.05, p = 0.005; domoic acid: R = 0.05, p = 0.002).
However, based on the low R-values, these findings likely do not represent a meaningful difference in diet. There were no significant differences in diet for California sea lions based on ANOSIM. Chi-squared and Fisher's Exact analyses identified several prey items significantly associated with the presence or concentration level of toxins found in the scat. In Steller sea lions, presence of saxitoxin was significantly different than expected the presence of non-juvenile salmonids, American shad Alosa sapidissima, walleye pollock, and skates (Table 6). When shad and pollock were present in the scat, saxitoxin presence was significantly higher than expected (Table 6). Saxitoxin was detected in every scat with pollock prey remains (n=18 scats). Presence of domoic acid was negatively associated with non-juvenile salmonids, but positively associated with spiny dogfish, Pacific sardine, and starry flounder Platichthys stellatus (Table 6). When sardine and starry flounder were present, domoic acid presence was significantly higher than expected (Table 6). In California sea lions, saxitoxin was positively associated with presence of non-juvenile salmonids, whereas scats with Northern anchovy were about five times more likely to have domoic acid compared to those without anchovy (Table 6).
When comparing presence of prey to specific toxin levels, the patterns were more complex. In Steller sea lions, if saxitoxin was detected in scats with rockfish, it was significantly more likely to be in low concentrations compared to other concentrations (Table 7). Saxitoxin appeared more frequently than expected in all concentration levels when non-juvenile salmonids were present, whereas skates were less frequent than expected in all concentration levels (Table 7). Spiny dogfish appeared most frequently associated with medium and high levels of saxitoxin (Table 7). Only walleye pollock was significantly different between saxitoxin concentration levels following Bonferroni adjustment (Table 8). Scats with pollock were significantly more likely to have medium and high levels of saxitoxin compared to low concentrations (Table 8). In spite of wide 63 confidence intervals, scats with sardine were significantly more likely to have high concentrations of domoic acid (Table 8). In California sea lions, Northern anchovy was detected with only low or medium concentrations of domoic acid (Table 8).  Table 7. Significant Pearson's Chi-squared tests (p<0.05) and p-values for 2 x 4 contingency tables comparing presence/absence of prey to levels of no, low (<20 ng g -1 ), med (20-50 ng g -1 ), and high (>50 ng g -1 ) concentrations of toxin in the scat. Chi-squared statistics (χ 2 , representing ∑ !!! ! ) and p-values for each 2 x 4 comparison are presented. Pearson's residuals are shown for presence of prey item:level of toxin, with residuals >|1.96| being significant (in bold). Positive residual values indicate higher than expected level of toxin and negative residuals indicate lower than expected level of toxin given the presence of each prey item.  Table 8. Significant Fisher's Exact tests (p<0.05) for 2 x 4 contingency, comparing presence/absence of prey to levels of no, low (<20 ng g -1 ), med (20-50 ng g -1 ), and high (>50 ng g -1 ) concentrations of toxin in the scat. Fisher's Exact p-values for the overall 2 x 4 contingency tables are presented. Post-hoc testing of all pairwise comparisons used a Bonferroni-adjusted alpha value to assess significance, where alpha was calculated as 0.05 divided by the number of pairwise comparisons. Significant 2 x 2 pairwise comparisons of presence of prey items at each toxin concentration level (e.g. presence of fish when low toxin levels in scat compared to presence of fish when no toxins in scat = Low:No) are presented by p-values with odds ratio and 95% confidence interval (CI). Bolded values indicate significance after Bonferroni-adjustment.  were only present at two southern beaches, Second Beach and Kalaloch Beach ( Figure   10). Domoic acid peaked in the fall of both 2011 and 2012, detected in concentrations above 1000 ng g -1 in 24 samples and detected in a single sample of California mussels from Makah Bay at a concentration <1000 ng g -1 (Figure 11).
To compare saxitoxin in sea lion scat to concentrations in bivalves, I overlaid plots on a map of the study area showing scat concentrations grouped by concentration level and toxin concentration in bivalve samples taken three weeks before and after the scat sample date (Appendix 1). In general, sea lion scat concentrations did not follow the same concentration trend as the bivalves. For example, in periods of very high toxins in bivalves, such as in July and August 2012 with levels measured as high as 6,000-12,000 ng g -1 of saxitoxin, the majority of sea lion scats did not have detectable levels (Appendix 1, Figure O). However, in September of that year, higher concentrations were observed in California sea lion scats (Appendix 1, Figure Q). Conversely, in April and May 2011, when most Steller sea lion scats had detectable levels of toxins and many in the ranges of 50-100 ng g -1 or >200 ng g -1 , bivalves had no to very low (<380 ng g -1 ) concentrations of 69 toxin (Appendix 1, Figures B and C). In several periods (e.g. February and March 2012), sea lions had detectable levels of saxitoxin (as high as 50-100 ng g -1 ), while no toxin was detected in nearshore bivalves (Appendix 1, Figures J and K).

Toxin Prevalence by Season, Year, and Location
This is the first study to document that marine mammals can be exposed to algal toxins year-round. Although Steller sea lions Eumetopias jubatus and California sea lions Zalophus californianus on the outer coast of Washington were more often exposed to saxitoxin than to domoic acid, both toxins were detected in scats in every season and month of the year. This last finding suggests that both saxitoxin and domoic acid were either retained in the food web for long periods after blooms  or that the toxins were transferred from the benthic or pelagic food webs through sediments or dormant cysts (Vigilant & Silver 2007, Jester et al. 2009, Sekula-Wood et al. 2011.

Given the bloom dynamics of Pseudo-nitzschia and Alexandrium in Washington
State, it is unlikely that there were active algal blooms in the winter when toxins were detected in sea lion scat. Harmful levels of domoic acid and saxitoxin may occur in Washington shellfish in late fall (e.g. November, Trainer 2002), while blooms in other areas may occur as late as December (Bates et al. 1998). On the Washington coast, Pseudo-nitzschia blooms are associated with the seasonal Juan de Fuca Eddy, which is formed in spring and dissipates in the fall (Hickey & Banas 2003). Alexandrium blooms require warming water temperatures typically beginning in the spring (around 13˚ C, Nishitani & Chew 1984). Thus, active blooms are not likely to occur in the winter.
Based on studies on domoic acid and saxitoxin retention in fish and mammals, it is unlikely that either the fish or the sea lions themselves were retaining the toxins for 74 long periods after the fall algal blooms. Domoic acid administered through oral gavage to northern anchovy Engraulis mordax and coho salmon Oncorhynchus kisutch was cleared from the digestive tracts within several days of exposure, but remained in some tissues for up to one week (Lefebvre et al. 2001(Lefebvre et al. , 2007. Paralytic shellfish toxins (specifically Nsulfocarbamoyl-11-hydroxysulfate toxins) were retained in fish as long as two weeks after exposure (Kwong et al. 2006). Conversely, some species of cephalopod appear to hold domoic acid for several months after algal blooms (Lage et al. 2012, Lopes et al. 2013; however, in my study, only two winter scat samples containing cephalopod remains also contained toxin (saxitoxin only). Although clearance rates of domoic acid and saxitoxin have not been studied directly in sea lions, in other mammal species clearance of these toxins typically occurs within 24 -48 hours (Truelove et al. 1997, Andrinolo et al. 1999, Wittmaack et al. 2015. Therefore, it is unlikely that the toxins were retained in the prey or sea lion tissues for extended periods.
It is more likely that detection of domoic acid and saxitoxin in scat year-round was due to transfer from the benthic to the pelagic food web. Both domoic acid , Vigilant & Silver 2007, Kvitek et al. 2008) and saxitoxin  have been detected in several benthic flatfish species documented in pinniped diet (Riemer et al. 2011. Several pathways for transfer of these toxins to the benthos have been proposed. Pseudo-nitzschia cells and particulate domoic acid (domoic acid within the cells) sink to the bottom where they accumulate in sediments or degrade and release domoic acid (Vigilant & Silver 2007, Sekula-Wood et al. 2011. Particulate domoic acid may also reach the benthos via fish or invertebrate fecal pellets (Vigilant & Silver 2007). Fish may then ingest domoic acid from feeding on benthic or epibenthic invertebrates (Vigilant & Silver 2007). Similarly, it is likely that resuspended Alexandrium cysts are re-incorporated into the marine food web leading to trophic transfer to top predators . Winter saxitoxin toxicity in shellfish is primarily attributed to cysts, known to contain high concentrations of saxitoxin (Schwinghamer et al. 1994, Wyatt & Jenkinson 1997, Bricelj & Shumway 1998). Cysts can be resuspended into the water column by wave action of winter storms (Kirn et al. 2005, Butman et al. 2014, Feifel et al. 2015. The overlap of deposited cells or cyst hotspots with the prey and foraging areas of marine mammals likely leads to winter saxitoxin exposure.
The exact pathways of benthic to pelagic food web transfer of toxins, whether from sediments or from resuspended cysts and plankton, are unknown in this study.
Consumption of flatfishes (order Pleuronectiformes) dropped sharply in the winter (Table   7), ruling out a strictly benthic feeding pathway. The most common prey items in winter scats containing saxitoxin were rockfish Sebastes spp. (22/31 scats) and non-juvenile salmon Oncorhynchus spp. (19/31 scats). Although about half of scats (16/31 scats) also contained clupeids (primarily unidentified or Pacific herring Clupea pallasii), herring and other clupeids typically fast in the winter (Paul et al. 1998, Foy & Paul 2011. The most common prey items in winter scats containing domoic acid (n = 5) were salmon, skates (Rajidae family), Pacific spiny dogfish Squalus suckleyi, and Pacific herring. Given that many of these species feed in the pelagic zone (Eschmeyer et al. 1996, Love 1996 and that the primary toxin detected in winter in this study was saxitoxin, it is likely that toxin exposure occurred through the resuspension of Alexandrium cysts into the water column (and thus into the marine food web) due to winter storms and wave action.  propose that salmon and rockfish feeding on krill at the surface or at depth may transport domoic acid up the food chain. Rockfish and salmon recovered in sea lion scats cannot be easily identified to species without otoliths (which are uncommonly recovered) or through molecular genetic techniques (Tollit et al. 2003, Purcell et al. 2004. However, it is most likely that the salmon in these scats were chinook Oncorhynchus tshawytscha or coho Oncorhynchus kisutch, which are the primary species caught in winter fisheries (Pressey 1953, Erickson & Pikitch 1994.
Based on the presence of other schooling fish (i.e. salmon, dogfish, and herring;Love 1996), the rockfish in these scats may also represent pelagic schooling species such as yellowtail rockfish Sebastes flavidus or widow rockfish Sebastes entomelas, which are common on the Washington coast (Keller et al. 2007). While the diets of salmon and rockfish differ by species, age class, and season, both feed on a variety of copepods, amphipods, and euphausiids as well as on juvenile and adult planktivorous fishes (Prakash 1962, Manzer 1968, Brodeur & Pearcy 1984, Brodeur et al. 1987. Thus, it is not possible based on this study alone to confirm by which of many possible pathways these fish acquired and transferred the toxins up the food chain to the sea lions. In Steller sea lions, domoic acid was more consistently found at high concentrations in the summer and at the Carroll/Sea Lion Rock complex ( Figure 6). The latter finding was somewhat surprising given the close proximity of Tatoosh Island to the Juan de Fuca Eddy. Domoic acid is detected in the Juan de Fuca Eddy throughout the summer, with the highest concentrations both in the eddy and in nearshore shellfish typically occurring in the fall (Trainer 2002, Trainer et al. 2009). If Steller sea lions fed in the vicinity of the eddy during the summer, then domoic acid would be detected earlier in 77 scats than in nearshore shellfish. Although it is unknown to what extent sea lions at either haulout foraged in or near the eddy, it is possible that the release of eddy waters to the southeast (MacFadyen et al. 2005, 2008, Trainer et al. 2009) allowed for domoic acid to enter the food chain closer to Carroll/Sea Lion Rock.
In contrast, there was no consistent season when saxitoxin was present in scats and there was no evidence that either haulout complex had consistently greater saxitoxin exposure ( Figure 6). This lack of pattern likely relates to differences in both bloom dynamics and sea lion diet. Based on concentrations in nearshore bivalves over the same time period, Alexandrium blooms appeared higher in the summer and fall in year 2 compared to year 1 (Figure 9), which could, in part, account for the differences in saxitoxin prevalence by season and location. However, in general there were more detections and higher concentrations of saxitoxin in the scats in year 1 compared to year 2 (Figures 3 and 4). Further, there was no bloom apparent in nearshore shellfish in spring of year 1 (Figure 9), suggesting the occurrence of a bloom outside of the sampled area or the existence of another factor, such as diet (discussed in detail below), may have been a greater contributor to exposure.

Diet Influences on Toxin Prevalence
Based on overall diet analyses using non-metric multidimensional scaling (NMDS), there was no apparent separation in diet when considering presence/absence of toxin (Figure 7) or toxin concentrations (Figure 8). Although the NMDS did not differentiate between season, year, or location, all of which may impact diet (Sinclair & Zeppelin 2002, McKenzie & Wynne 2008, Riemer et al. 2011, the lack of a convergent solution and lack of spatial groups in the NMDS plots (showing best solution after 20 iterations) is indicative of the diversity of sea lion diet (Tables 7 and 8).
Most compelling in this study is evidence that both walleye pollock (hereafter pollock) and American shad (hereafter shad) may be vectors of saxitoxin exposure in Steller sea lions (Tables 6 and 8). Saxitoxin was detected in all scats containing pollock (n = 18 scats) and in most scats containing shad (13/18 scats). Because both species had relatively low frequency of occurrence in sea lion diet overall (Table 7), this finding requires some scrutiny. The occurrence of pollock is primarily traced to a single collection day (18 May 2011) at Tatoosh Island complex, where 17/17 scats were positive for saxitoxin; 15 of these contained pollock. The two remaining scats without pollock contained unidentified codfish bones, which could have also been pollock bones that were either degraded by digestion or were structures that could not be identified to species. The majority of these scats also contained spiny dogfish (14/15), however only 46 % of all scats containing dogfish had detectable saxitoxin. Six of these 15 scats also contained Pacific herring (1 scat) or unidentified clupeids (5 scats). Pollock was also found in three other scat samples in this study and saxitoxin was detected in all three samples. Scats with pollock present were significantly more likely to have medium (20-50 ng g -1 ) or high (>50 ng g -1 ) concentrations of saxitoxin. Scats with pollock contained two of the highest concentrations of saxitoxin measured (214.4 ng g -1 and 273.6 ng g -1 ) and were the driver of the high prevalence and concentrations of saxitoxin in spring 2011 ( Figure 6).
If pollock are indeed a vector of saxitoxin (or other algal toxins) to Steller sea lions, this could be a significant problem for the western DPS sea lions that rely on this fish in their diet. Pollock are a mid-water to benthic species that eat a variety of zooplankton and other fishes (Eschmeyer et al. 1996, Love 1996, Brodeur 1998 and are a common prey species in Steller sea lion diet in the eastern Aleutian Islands, Gulf of Alaska, and southeast Alaska , Sinclair & Zeppelin 2002, Bredesen et al. 2006). The low occurrence of pollock in scats in this study likely relates to their overall distribution, which is highest in the Bering Sea and Gulf of Alaska and lower along the coast of Washington and Oregon (Bredesen et al. 2006). Pollock were caught in low numbers in bottom trawl surveys on the Washington coast (Keller et al. 2007) and are caught infrequently as bycatch in tribal bottom trawl fisheries primarily around Swiftsure Bank (pers. comm. Joe Petersen 1 ), a high spot offshore the Strait of Juan de Fuca. The proximity of Tatoosh Island to this location provides a likely explanation for why the majority of scats with pollock were from the Tatoosh Island complex. Although only low concentrations of domoic acid and saxitoxin have been documented in stranded Steller sea lions in Alaska (Lefebvre et al. 2016), further assessment of algal toxins levels in free-ranging sea lions is warranted. Given that several of the animals tested by Lefebvre et al. (2016) were pups or aborted fetuses and that stranded animals may not be foraging normally (Jensen 2015), the concentrations reported in those stranded sea lions may not represent exposure to live animals.
It is possible that shad also acted as a vector of saxitoxin to Steller sea lions in this study. Scats with shad (n = 18) were collected on a total of 10 collection days, in spring, fall, and winter of both years. These scats contained one to four additional prey items from six different family groups, therefore it cannot be ruled out that the presence of toxin was associated with another prey item. American shad eat zooplankton and small fish (Love 1996) that may be directly ingesting toxic phytoplankton, but neither toxin has previously been documented in shad. Scats with shad were associated with low concentrations of the toxin (<20 ng g -1 ), suggesting that although they may have been vectors of the toxin, they resulted in relatively low toxin exposure.
Possible vectors identified for domoic acid were Pacific sardine Sardinops sagax and starry flounder Platichthys stellatus in Steller sea lions and northern anchovy in California sea lions (Tables 6 and 8). Domoic acid was detected in 57 % of Steller sea lion scats containing sardine (17/36 scats). These scats were collected in summer and fall of both years and contained one to seven other prey items. Scats with sardine included several of the highest concentrations of domoic acid detected in this study, including one sample with 423 ng g -1 . Given that sardine is a known vector of domoic acid and cause of marine mammal mortalities (Scholin et al. 2000) and that sardine may feed directly on the toxin-containing diatoms (Costa & Garrido 2004), it is likely that this species was, at least in part, responsible for the detection of domoic acid in these scats.
Another possible vector for domoic acid was starry flounder, with 36% of scats containing starry flounder (13/36) also containing detectable levels of domoic acid. These scats contained one to seven other prey, most commonly skates and dogfish. Starry flounder are typically found nearshore in shallow waters in the spring and summer and 81 move offshore in the winter, primarily feeding on benthic mollusks, crustaceans, and fishes (Eschmeyer et al. 1983, Love 1996. These fish were found almost exclusively in scats from the Carroll/Sea Lion Rock complex, collected in spring, summer, and fall of both years. Given that starry flounder are not likely to be found in the offshore areas of the Juan de Fuca Eddy during the summer when domoic acid is most prevalent, this may be further evidence that waters escaping the eddy to the southeast influence the benthic marine food web in nearshore environments. Northern anchovy was likely a vector of domoic acid in California sea lions, with domoic acid detected in 6 of 14 samples that contained anchovy. Domoic acid was detected in scats primarily in November 2011 (5/6 scats), but scats contained several other prey species including herring, sardine, and shad. Being a known vector of domoic acid  it is possible that anchovy did expose sea lions to domoic acid in this case. However, the small sample size suggests that this species did not provide significant exposure, and the presence of other planktivorous prey species make it impossible to know whether one or more prey species contained the toxin.
Other prey species that were identified as possible vectors of toxin exposure, though less convincingly, include non-juvenile salmon and spiny dogfish (Tables 6 and   7). In Steller sea lions, about 53% of scats containing non-juvenile salmon also had detectable levels of saxitoxin, with the highest prevalence in summer and fall. In California sea lions, only 43% of scats containing non-juvenile salmon also contained saxitoxin. Similarly, domoic acid in Steller sea lion scats was positively associated with presence of spiny dogfish, although only 24% of scats (26/108 scat) containing dogfish were positive for the toxin. These scats were collected in all seasons of both years and contained one to seven other prey items, most commonly skates (17/26 scats) and clupeids (19/26 scats). Although algal toxin accumulation has been documented in salmon (Lefebvre et al. 2007, Sephton et al. 2007), toxins in elasmobranch species have not been reported in the wild (Vigilant & Silver 2007). Further, leopard sharks Triakis semifasciata given oral doses of domoic acid showed low retention of the toxin (Schaffer et al. 2006). Given the variety of other prey species in these scats, it is difficult to confirm that either non-juvenile salmon or dogfish were specifically related to the presence of toxin. The diversity of prey associated with toxins in this study, including planktivorous forage fish, benthic flatfish, and higher-tropic pelagic species, suggests that a variety of pathways lead to HAB toxin exposure in these top-predators.

Comparison to Nearshore Bivalves
In this and previous studies , Torres de la Riva et al. 2009), sea lions appear to act as indicators of offshore blooms, such as those occurring in the Juan de Fuca Eddy, before they are apparent in the nearshore waters (Figures 9 and 11).
Therefore, nearshore monitoring of phytoplankton and particulate toxins likely underrepresents toxin exposure in marine mammals. Other studies have found evidence for co-occurrence of the toxins in marine mammals and in bivalves (Fire et al. 2010) and for a lack of correlation between the two (Scholin et al. 2000, Bargu et al. 2010). One study found significant correlations between marine mammal strandings and toxic phytoplankton blooms at several time scales, both before and after Pseudo-nitzschia blooms were otherwise detected (Torres de la Riva et al. 2009). In my study, the detection of both saxitoxin and domoic acid in nearshore shellfish peaked in summer and fall of both years ( Figure 10). Although the comparison of domoic acid in scats to that in bivalves was limited, sea lions appeared to carry the toxin somewhat prior to the spikes seen in nearshore shellfish (Figure 11). This precedence likely indicates that the sea lions were either foraging in areas of blooms, such as the Juan de Fuca Eddy, prior to those waters making it nearshore in the fall, or could indicate that their prey species travelled through or foraged in these areas.
The comparison between saxitoxin in bivalves and sea lions was more complex.
High concentrations of saxitoxin were detected in sea lion scat (>200 ng g -1 ) prior to concentrations in bivalves reaching levels above the regulatory limit, such as in spring 2011 (Appendix 1, Figure C). Similarly, in periods of relatively high concentrations in bivalves, such as in the summer and fall (Appendix 1, Figures C and O), some scats had medium and high concentrations (up to 50-100 ng g -1 ), but most scats were below detection. The exception was California sea lions, which experienced high levels of saxitoxin (>200 ng g -1 ) at the same time as very high concentrations (~12,000 ng g -1 ) were detected in bivalves (Appendix 1, Figure Q). Of particular interest is the occurrence of saxitoxin in sea lion scat when no to low (<380 ng g -1 ) saxitoxin was detected in bivalves, as seen in winter and spring (e.g. Appendix 1, Figures J-L), perhaps further confirming that winter toxin exposure in sea lions is through benthic pathways rather than active blooms. If this exposure resulted from Alexandrium cysts being resuspended into the water column, it is possible that this occurred outside of areas that are presently 84 sampled for bivalves, and thus was not detected in the bivalves, or it is possible the cysts re-entered the food web through another pathway.

Study Limitations
Despite many compelling findings, there are several limitations of diet analyses including how prey are reported in the diet, how the different prey species are digested, and how the toxins are metabolized. Frequency of occurrence (FO) is a commonly used metric for describing diet in pinnipeds (Lance et al. 2001), but has several limitations. FO describes the presence or absence of a prey item without enumerating the number of fish eaten or estimating volume or biomass of eaten prey. Scat analysis in general has known biases related to identifying prey structures, degradation of bones in digestion, and different passage and recovery rates of hard parts of fish ingested (Orr & Harvey 2001, Cottrell & Trites 2002, Tollit et al. 2003, 2007. In general, smaller fish species may pass through digestion more quickly compared to larger prey that may take several days for bones to pass (Orr & Harvey 2001, Tollit et al. 2007).
Because both saxitoxin and domoic acid are typically flushed from the digestive system quickly, it is likely that toxin in scats are from a recent feeding event. Although this study incorporated the use of all prey structures, which improves the diet analysis result (Lance et al. 2001, Browne et al. 2002, Cottrell & Trites 2002, it is possible that the hard parts in a single scat do not represent all fish eaten or do not represent all fish from the most recent meal. Given that multiple fish species have been documented as having domoic acid and saxitoxin and that sea lion diet is very diverse (McKenzie & Wynne 2008, Riemer et al. 2011, it is difficult to infer which prey item or items actually held the toxin. My study incorporated anonymous scats from mixed demographic haulout sites and thus did not account for differences in diet and foraging behavior by age or sex (Merrick & Loughlin 1997, Trites & Calkins 2008, Orr et al. 2011, which in turn can affect exposure to HAB toxins (Silvagni et al. 2005. Lastly, there is no literature on what effects weathering might have on the toxin content in scats deposited on these rock and beach haulout sites. Therefore, the toxin concentrations in this study are assumed to represent the full amount at the time of deposition.

Consequences of Toxin Exposure
Despite evidence of year-round exposure, toxin concentrations measured in this study are not likely to have caused acute illness. Marine mammal strandings in Washington State due to acute HAB toxicity have not been yet confirmed. Northern fur seals Callorhinus ursinus stranded in Washington and Oregon in the 1980's exhibited clinical signs of domoic acid toxicity, however these cases occurred before the toxin was recognized to cause health effects in marine mammals and was not confirmed as the cause of the strandings (Lefebvre et al. 2010). Stranded Steller sea lions in Washington have been tested for domoic acid, but only low concentrations have been found (Wiles 2015). Low levels of saxitoxin were documented previously in Washington in three live sea otters (White et al. 2013), but testing for saxitoxin has not been done in other species.

86
Domoic acid concentrations in this study were well below those measured in scat from California sea lions and harbor seals exhibiting acute domoic acid toxicity (Lefebvre et al. 1999, McHuron et al. 2013), but were comparable to the lower range of concentrations in scat of California sea lions with acute and chronic symptoms reported by Goldstein et al. (2008). The concentrations were higher than those measured in stranded, harvested, and live-captured pinnipeds in Alaska (Lefebvre et al. 2016), but significantly lower than maximum concentrations in scat reported in stranded northern fur seals in Alaska (Lefebvre et al. 2010) and free-ranging harbor seals in Scotland .
It is not known what concentrations of saxitoxin cause mortality in marine mammals. In humpback whales and Mediterranean monk seals affected by saxitoxin, samples of feces and other body fluids were not available for comparison to this study.
Humpback whales were estimated to have ingested about 3,200 ng saxitoxin kg -1 body weight (Anderson & White 1989), which is much higher than any concentration measured in this study. Concentrations of saxitoxin in my study were higher than any reported in 13 marine mammals species sampled in Alaska (Lefebvre et al. 2016), but within the range of saxitoxin measured in feces of free-ranging and stranded harbor seals in Scotland  and in free-ranging North Atlantic right whales in the Bay of Fundy, Canada (Doucette et al. 2006).
Estimating the doses of toxin that cause toxicity in marine mammals is difficult and comparing scat concentrations to toxicity is complicated by the fact that feces does not necessarily represent the concentrations absorbed by the digestive tract. While concentrations in feces represent exposure, measuring toxins in urine or serum is more 87 indicative of the amounts metabolized (Fire et al. 2009, Lefebvre et al. 2010. In other studies, concentrations of domoic acid measured in feces were frequently orders of magnitude higher than those measured in the urine or serum (Scholin et al. 2000, Fire et al. 2009, 2010, Lefebvre et al. 2010, McHuron et al. 2013. Saxitoxin concentrations similar to those in this study were reported in the feces of harbor seals in Scotland; however, no saxitoxin was detected in the urine (Jensen 2015).
Whether the lack of detection in urine could be related to low toxin exposure and thus low absorption across the digestive tract was not discussed. Therefore, it is difficult to infer what urine concentrations, and thus metabolized concentrations, might have been present in the sea lions studied here.
The concentrations measured in feces in this study likely indicate very low concentrations metabolized into tissues. Although sea lions have a high digestion efficiency of fish (>90%, Rosen & Trites 2000), the assimilation efficiency of these HABs toxins is not known. Absorption of toxins across the digestive tract differs greatly between mammal species . In terrestrial mammal species, only about 4-7% of ingested domoic acid is absorbed across the digestive tract (Truelove et al. 1997). The amount of saxitoxin absorbed by the digestive tract has not been reported in mammals, however there is evidence in both humans (García et al. 2004) and whales (Doucette et al. 2006) of biotransformation of saxitoxin into other derivatives.
Although the concentrations measured in this study were low, massive coast-wide algal blooms occurred throughout the northern California Current in the spring and summer of 2015, leading to high concentrations of both domoic acid and saxitoxin and to coast-wide shellfish closures (Peterson et al. 2013). The results of my study could have 88 been drastically different had they included a year with large algal blooms such as those observed in 2015 and could have lead to different detection rates of the toxins and potentially health-threatening levels. Future blooms of similar magnitude could have profound effects on the health of sea lions on the Washington coast.

Conclusions
This study presents evidence of chronic low-level exposure to saxitoxin and domoic acid in Steller and California sea lions. While chronic exposure to domoic acid has been well documented in California sea lions , it is unclear what concentrations induce chronic toxicity and how those concentrations compare to the concentrations to which sea lions in Washington may be exposed. In addition to epilepticlike symptoms, one consequence of chronic exposure to domoic acid is compromised navigational ability, which can affect foraging and migration behavior (Thomas et al. 2010, Cook et al. 2015. Chronic saxitoxin toxicity has not been documented, although several authors have proposed possible effects. Durbin et al. (2002) suggest that chronic exposure to saxitoxin may affect diving capability or result in greater susceptibility to disease. Bogomolni et al. (2016) suggest that saxitoxin may increase the risk of morbillivirus and phocine distemper virus in harbor seals.
What effects either chronic or acute algal toxin exposure may have at the population level remain unknown. While Jensen et al. (2015) explored the possibility that HAB toxins may be related to the decline in harbor seals in Scotland, they observed no 89 apparent toxicity in any of the animals tested. Despite the rising number of strandings of California sea lions due to domoic acid toxicity , the population has been steadily increasing (Carretta et al. 2014). Given that female California sea lions are disproportionally affected by acute domoic acid toxicity (Silvagni et al. 2005) and that domoic acid can cause reproductive failure in California sea lions (Brodie et al. 2006), this toxin puts a more sensitive part of the population at risk.
While the reasons for the decline in western DPS Steller sea lions remain unknown (NRC 2003), it is evident that both domoic acid and saxitoxin are present in the Arctic food web and could play a role (Lefebvre et al. 2016). Given the small sample size in this study, the implication that a dominant prey resource of the western DPS population (i.e. walleye pollock) may be a vector of saxitoxin exposure warrants further study. This study also found evidence that prey with relatively low occurrence in the diet may act as vectors of significant algal toxin transfer up the food chain. Therefore, generalist predators with a more diverse diet may not have any respite from exposure to, or the effects of, marine algal toxins as compared to predators that specialize. Sampling at a finer timescale (e.g. weekly or monthly) may be necessary for understanding the role of infrequent prey species in transfer of algal toxins to top predators. This study confirms for the first time that marine mammals can be exposed to algal toxins through their prey outside of active algal bloom time periods and that benthic to pelagic food web transfer of precipitated cells and dormant cysts is the most likely cause.
Moore SK, Trainer VL, Mantua NJ, Parker MS, Laws EA, Backer LC, Fleming LE (2008) Impacts of climate variability and future climate change on harmful algal blooms and human health. Environ Health 7 and California sea lions Zalophus californianus (gray) are summarized by concentration level (left) as having no toxin, 0-20 ng g -1 , 20-50 ng g -1 , 50-100 ng g -1 , 100-200 ng g -1 and > 200 ng g -1 . Scatterplots show concentrations of saxitoxin (ng g -1 ) by sample date, with individual dots representing the concentration measured in 100 g of bivalve tissue. Time periods are defined as three weeks before and three weeks after scat sampling dates. Dates with * indicate time periods where two scat collection occurred within one week and the time period displayed was lengthened. Note that some dates are repeated in subsequent graphs. Appendix 1 cont. Saxitoxin concentration (ng g -1 ) by site in sea lions (barplots) and nearshore bivalves (scatterplots). Scats from Steller sea lions (black) and California sea lions (gray) are summarized by concentration level (left). Scatterplots show concentrations of saxitoxin (ng g -1 ) by sample date, with individual dots representing the concentration measured in 100 g of bivalve tissue. 113 Appendix 1 cont. Saxitoxin concentration (ng g -1 ) by site in sea lions (barplots) and nearshore bivalves (scatterplots). Scats from Steller sea lions (black) and California sea lions (gray) are summarized by concentration level (left). Scatterplots show concentrations of saxitoxin (ng g -1 ) by sample date, with individual dots representing the concentration measured in 100 g of bivalve tissue. 114 Appendix 1 cont. Saxitoxin concentration (ng g -1 ) by site in sea lions (barplots) and nearshore bivalves (scatterplots). Scats from Steller sea lions (black) and California sea lions (gray) are summarized by concentration level (left). Scatterplots show concentrations of saxitoxin (ng g -1 ) by sample date, with individual dots representing the concentration measured in 100 g of bivalve tissue. 116 Appendix 1 cont. Saxitoxin concentration (ng g -1 ) by site in sea lions (barplots) and nearshore bivalves (scatterplots). Scats from Steller sea lions (black) and California sea lions (gray) are summarized by concentration level (left). Scatterplots show concentrations of saxitoxin (ng g -1 ) by sample date, with individual dots representing the concentration measured in 100 g of bivalve tissue. 126 Appendix 1 cont. Saxitoxin concentration (ng g -1 ) by site in sea lions (barplots) and nearshore bivalves (scatterplots). Scats from Steller sea lions (black) and California sea lions (gray) are summarized by concentration level (left). Scatterplots show concentrations of saxitoxin (ng g -1 ) by sample date, with individual dots representing the concentration measured in 100 g of bivalve tissue.