Foraging behaviour of the South American sea lion (Otaria byronia) in two disparate ecosystems assessed through blubber fatty acid analysis

Fatty acids have been widely used as trophic biomarkers in marine mammals. However, for the South American sea lion, the most abundant otariid in the eastern South Pacific, there is no information about blubber fatty acids and their link to diet. Here, we compare fatty acid profiles of sea lions from two distinct oceanographic regions in northern and southern Chile. Their fatty acids vary greatly between regions, suggesting dietary differences at a spatial scale. The fatty acid C22:6ω3 was more abundant in sea lions from the northern region, likely associated with consumption of anchovy, cephalopods, and crustaceans, which are rich in that fatty acid, and have been reported as their main prey items. Sea lions from the southern region were richer in C22:1 and C20:1, characteristic of teleost fish, suggesting a piscivorous diet. Males displayed a more diverse fatty acid composition than females, suggesting a wider trophic niche. Few individual sea lions within the southern region had unusually high levels of C18:2ω6, commonly found in terrestrial environments. This suggests consumption of farmed salmon, whose diet is usually based on terrestrial sources. This demonstrates how human intervention is being reflected in the tissues of a top predator in a natural environment.


Results
A total of 19 fatty acids were identified consistently in amounts higher than 0.5% in the blubber of South American sea lions (Table 1). Overall, sea lions displayed great individual variation in their blubber fatty acids. The outer blubber layer of sea lions was dominated by monounsaturated fatty acids both in the northern (43.3 ± 10.4%) and the southern region (51.6 ± 14.3%). Saturated fatty acids were the second most abundant (27.7 ± 9.6% and 25.7 ± 6.8% in the northern and southern regions, respectively) followed by polyunsaturated fatty acids which were present in higher proportions in the northern (16.6 ± 12.0%) compared to the southern region (10.5 ± 8.5%). The six most abundant fatty acids, in decreasing order, were: C18:1ω9, C16:0, C16:1ω7, C18:0, C22:6ω3 and C18:1ω7, which accounted for 69.2% and 70.5% of the total fatty acids, in the northern and southern regions, respectively.
Both region (PERMANOVA, F 1 = 19.27, P = 0.008; Fig. 2) and sex within each region (PERMANOVA, F 2 = 3.37, P = 0.008; Fig. 3) influenced the overall fatty acid composition of sea lions. Region explained 17% of the variation whereas sex only explained 6%. SIMPER indicated that the average dissimilarity between northern and southern regions was 24%, and that the fatty acids most responsible for these differences were: C18:1ω9, which was more abundant in sea lions from the southern region, C22:6ω3, which was higher in the northern region, and C18:0, C16:1ω7, C16:0 and C14:0, most of them slightly higher in sea lions from the northern region (Tables 1  and 2). These six fatty acids explained 69% of the variance.

Discussion
To our knowledge, this is the first study that describes the fatty acid composition of South American sea lions from different marine ecosystems, including sex classes, and allowing inferences to be made about their dietary patterns. This new approach provides insights into the general foraging behaviour of these sea lions, and not only their short-term diet. This qualitative analysis of fatty acids in a top predator allows us to infer energy flows within the ecosystem, promoting further research based on these trophic biomarkers.    www.nature.com/scientificreports www.nature.com/scientificreports/ South American sea lions have high proportions of C18:1ω9, C16:0, C16:1ω7 and C18:1ω7, consistent with the main fatty acids found in other marine mammals 32,55,56 . These fatty acids have a mixed origin; they can be synthesised endogenously but are also influenced by diet 46 . Conversely, high amounts of C22:6ω3, or docosahexaenoic acid (DHA), are due to dietary influence, as this type of polyunsaturated fatty acid is synthesised by primary producers 25 . Thus, the abundance of DHA in high trophic levels is associated to consumption of prey rich in DHA and selective retention of this fatty acid by consumers 36 . Although DHA is abundant in the marine environment,  Table 3. Contribution of each fatty acid (%) to the dissimilarity between males and females within the southern region, in descending order. P values in bold indicate significant differences between sexes.
South American sea lions have high (e.g., ~7%) amounts of C18:0 (Table 1) compared to other carnivorous marine mammals (e.g., where values rarely overcome 3% 30,56,57,61,63,64 ). This fatty acid has a mixed origin, where relatively large contributions are from both biosynthesis as well as diet 46 . A high proportion of C18:0 could be a characteristic of the species, where males have higher proportion than females, or may reflect the abundance of this fatty acid in the ecosystem along the eastern South Pacific coast. Only a few studies on fatty acids in the marine ecosystem have been conducted in the area, which include two species of seaweed 65 and a few species of fish 66 . Seaweeds contain around 8% of C18:0 and fish range between ~6% and ~8%. Thus, since this fatty acid is abundant in organisms of low trophic levels, and due to the little modification of fatty acids passed up the food chain, this top predator should reflect the abundance of this fatty acid in the ecosystem.
As expected, sea lions from the northern region differ in their fatty acid composition compared to those from the southern region as these populations feed within very different oceanographic systems of the eastern South Pacific. Although fatty acid data of potential prey are necessary to establish the diet of a predator in a quantitative manner, the dissimilarities in fatty acids, particularly in those of dietary origin, suggest differences in dietary preferences at this spatial scale. Our findings of this regional dietary difference support previous studies based on stable isotopes that show a great variation in diet composition between northern and southern populations of South American sea lions in Chile 4 . One of the main drivers of segregation was C18:1ω9, which was more abundant in sea lions from the southern region. High proportions of C18:1ω9 are characteristic of many fish species, in particular when C20:1ω9 and C22:1ω11 are also abundant 67 . These other two fatty acids, C20:1ω9 and C22:1ω11, were also identified as responsible of the segregation of sea lions' fatty acids due to their higher values in the southern region. Copepods synthesise substantial amounts of isomers of C22:1 and C20:1 68 . These fatty acids are conservatively incorporated into the tissues of consumers, in particular fish. Thus, they are characteristic of many teleost fish, along with the fatty acid C18:1ω9 67 , which indicates that the consumption of fish is potentially greater in sea lions from the southern region. This is confirmed by previous studies in the area that report that their diet is dominated by demersal fish 4,6 .
DHA was the second main contributor to dissimilarity between regions, where the proportion of DHA in South American sea lions from the northern region doubled that of sea lions from the southern region. This ω3 fatty acid is a good dietary indicator, which suggests that sea lions from the northern region are feeding on prey rich in DHA, although with great individual variability. Stomach and scat content studies in northern Chile have indicated that the Patagonian squid Loligo gahi, the anchovy Engraulis ringens and the squat lobster Pleuroncodes monodon are the main prey species 54,69 . This coincides with the abundance of DHA in the blubber of sea lions, as cephalopods, crustaceans and fish are rich in DHA, with cephalopods having the highest proportions of DHA within the marine environment 36 . Unlike most fish species, anchovies contain only trace amounts of C22:1 and C20:1 isomers 70 , which explains why sea lions from the northern region, although incorporating fish in their diet, have lower amounts of these isomers compared to sea lions from the south.
The fatty acids C16:ω7 and DHA are also associated to pelagic ecosystems 38 . The abundance of both fatty acids in the northern sea lions suggests that they feed in offshore pelagic habitats. This is consistent with the marine ecosystem off northern Chile where the narrow continental shelf (~200 m) 52 provides limited opportunity for seafloor benthic foraging. Conversely, sea lions from the southern region inhabit the waters of the inner channel ( Fig. 1), which offers more opportunities for benthic feeding. Even if sea lions travelled to more oceanic waters off the outer coast, the continental shelf in this southern area is wider (50 km) than in the northern region 52 ; therefore, a more benthic foraging would be expected and has, in fact, been reported by Sepúlveda et al. 71 .
Although in the northern region the difference in fatty acid compositions between males and females was not significant, in both regions the pattern is similar: the fatty acids of males are more differentiated from each other, whereas for females the fatty acid composition is more similar (Fig. 3). This would indicate that males have a wider trophic niche than females, and that each male individual have a unique diet that is not very similar to that of its counterparts. Similarly, using stable isotope analysis Sepúlveda et al. 6 showed that subadult and adult males have a broader isotopic niche width than females. This coincides with reports of males travelling longer distances and diving deeper than females 12 , thus, exploring greater areas and increasing the opportunities of having different diets.
Males and females from the northern region did not display differences in their overall fatty acid composition or dietary fatty acids. Potentially, the fact that the study area is larger than that covered in the southern region, implies that there are other sources of variation within the region not accounted for. Further, the number of males and females is unbalanced, which complicates statistical analyses.
In the southern region, sex classes were different in overall fatty acid composition (Fig. 3B) and in proportions of the dietary isomers C20:1 and C22:1 (Fig. 6). According to stomach content studies conducted in the Atlantic coast, males consume a broader trophic spectrum than females 8 . In southern Chile, Sepúlveda et al. 6 found that both males and females only differed slightly in their isotopic values, where both sexes fed mainly on demersal fish, but adult females consumed pelagic fish as their second favourite prey whereas males preferred bentho-pelagic fish. Since fatty acids have greater power of distinction among prey species than stable isotopes 72 ; this segregation between sex classes found in this study likely reflects subtle differences in diet. Unfortunately, without fatty acid information of potential prey species, it is difficult to infer whether females are preying more heavily on certain fish species (e.g., richer in C20:1 and C22:1) than males.
Although dietary fatty acids are markedly different between regions, as expected for dissimilar ecosystems, variations among colonies within each region were not evident (Fig. 4). Potentially, individual differences are more important. Using satellite telemetry, Hevia 73 studied the foraging areas used by South American sea lions from the colony Patache, in northern Chile, and found that sea lions travelled long distances to forage, mainly parallel to the coastline, although a few individuals explored oceanic waters. Other studies on the species have reported segregation in foraging areas for lactating females, where some females are predominantly coastal in their foraging habits whereas others are more pelagic 12 . Similarly, Sepúlveda et al. 73 reported inter-individual differences in δ 13 C values in South American sea lions from southern Chile, suggesting different foraging areas, which matched satellite tracking data derived from the same individuals. This coincides with findings of Muñoz et al. 4 , who suggest that South American sea lions display individual trophic specialisation.
High levels of ω6 fatty acids are consistent with terrestrial and freshwater environments, whereas low levels are characteristic of marine ecosystems 68 . For example, when comparing freshwater and marine ringed seals, Pusa hispida, Käkelä et al. 74 found that the former had almost five times greater proportion of C20:4ω6 than the marine seal population. Similarly, Smith et al. 75 reported significantly higher proportions of C20:4ω6 and C18:2ω6 in freshwater harbour seals, Phoca vitulina, compared to their marine relatives. Other freshwater mammals such as some species of beavers and otters have levels of C18:2ω6 that range from ~10-30%, being sometimes the most abundant fatty acid in their fat depots 76,77 . Here, differences in ω6 values between sea lions from northern and southern regions are important for C18:2ω6 but not for C20:4ω6.
Although sea lions from both regions are marine animals, the southern study area receives an important fresh water discharge, as this area is within one of the most extensive fjord regions in the world 53 . Sea lions from the northern region have levels of C18:2ω6 within the normal range reported for other otariids, which rarely overcomes ~1.5% 56,57,[60][61][62] . Conversely, sea lions from the southern region have higher levels of C18:2ω6 (Table 1, Fig. 5), although with substantial inter-individual variation, where a few sea lions, especially males, had unusually high values of this fatty acid (Fig. 6). C18:2ω6 is characteristic of vascular plants and is considered dietary since their conversion from C18:1ω9 requires Δ12 desaturase enzymes found only in primary producers 78 . This suggests that there is individual specialisation, where most of the population feed on a wholly marine ecosystem, and some sea lions are exploiting prey from inland habitats. South American sea lions are known to enter river mouths and Atlantic coastal lagoons along their distribution zone 79 . In Chile, a very well-known example is a small resident colony that inhabits the Valdivia River 80 , whereas in other areas in southern Chile they have been reported to swim up the rivers to feed on feral salmon 73 . However, sea lions in this study were sampled in a marine area, which indicates that they do not live in the rivers permanently. Thus, an intermittent freshwater foraging would not entirely explain why these sea lion males have values of C18:2ω6 equal or higher than seals that spend their entire lives in freshwater habitats.
Another explanation to higher levels of ω6 fatty acids found in some sea lions in the southern region is the consumption of farmed salmon. This area is characterized by a high concentration of exotic salmon farms. Different studies have reported frequent attacks to these salmon farms by South American sea lions 10,81 and, through stable isotope analysis, farmed salmon has been identified as one of the main contributors to the diet of South American sea lions, especially for males 4,6,73 . The use of products such as vegetable oil, soybean, poultry and wheat [82][83][84] in pellets of farmed salmon contribute to an increase in their proportions of fatty acids of terrestrial origin, such as C18:2ω6. Thus, farmed salmon has been found to contain up to six times higher proportions of ω6 fatty acids compared to wild salmon 83 .
Although most male sea lions have C18:2ω6 values within the normal range reported for marine mammals, some of them have C18:2ω6 values comparable to freshwater seals, with one male having values similar to those found in terrestrial mammals (Fig. 7). The figure illustrates the average value of this fatty acid for farmed Atlantic salmon, Salmo salar, based on studies conducted in the early 1990s and 2000s 83,85,86 ; however, this value is potentially underestimated, as Sprague et al. 87 demonstrates that, between 2006 and 2015, the integration of terrestrial-derived ingredients in the diet of farmed salmon has led to a steady increase of terrestrial fatty acids in farmed Atlantic salmon from Scottish waters. Unfortunately, little is known about the fatty acid composition of farmed salmon in our study area.
Consequently, high levels of ω6 fatty acids in the blubber of sea lions are potentially related to heavy consumption of farmed salmon by a few individuals, since freshwater foraging would not explain values as high as those of terrestrial mammals. The consumption of farmed salmon is a foraging strategy adopted by only a part of the population 73 . For example, Muñoz et al. 4 , using stable isotope analysis, found that whereas salmon contributed more than 50% to the diet of a few South American sea lions, for others the contribution of salmon was insignificant. Similarly, using satellite tracking, Sepúlveda et al. 73 determined that of 8 sea lions tracked, only one spent more than 20% of the time within close proximity (1 km) of salmon farms whereas the other 7 spent between 0.1 and 8.5%. Additionally, the median contribution of farmed salmon to the diet of these 8 sea lions, estimated through hair stable isotopes, ranged from 8 to 48% 73 .
Here, all values of C18:2ω6 higher than 4% correspond to adult and subadult males. Coincidently, salmon consumption is a foraging strategy adopted mainly by subadult and adult males, the age/sex classes most frequently involved in interactions with salmon farms in the region 81 . This is the first study, to our knowledge, to demonstrate how human-derived food has introduced a foreign fatty acid into the tissues of a top predator living in a natural marine ecosystem. Whether these unusual fatty acid signatures are also present in organisms at lower trophic levels, is something that remains to be investigated.
Overall, the dietary inferences that we made based on blubber fatty acids of the South American sea lion concord with our current knowledge about its foraging habits; however, more accurate, and quantitative, results could be obtained if fatty acid data of potential prey species were available. In order to understand trophic interactions, further studies on fatty acid composition of different marine organisms including farmed and feral salmonids are needed, in particular in the eastern South Pacific.

conclusions
This study demonstrated that sea lions from different oceanographic regions have contrasting blubber fatty acids, which reflects variations in prey availability. Sea lions from the northern region prefer prey rich in C22:6ω3, like anchovy, cephalopods and crustaceans, whereas southern sea lions consume more fish, rich in C22:1ω11, C20:1ω9 and C18:1 ω9. Unusually high levels of C18:2ω6, a fatty acid abundant in terrestrial environments, in sea lions from the southern region would indicate consumption of farmed salmon and demonstrates how salmon farming is potentially introducing a foreign compound into a natural ecosystem.

Materials and Methods
Sample collection. South American sea lions (n = 97) were sampled in two main areas in the north and south of Chile, which from here on will be referred to as the northern and southern regions, respectively. The three northern study sites, located between 20°48'S and 23°06'S, were Patache, Punta (Pta) Lobos, and Mejillones. More than 2,000 km southwards, between 41°42'S and 42°28'S, the three southern study sites, La Arena, Chullec and Cahuelmó, are sea lion colonies within the inner sea of the Los Lagos Region of southern Chile (Fig. 1). This area is characterised by a high number of salmon farming installations. Field work was conducted during the austral autumn of 2015 (Table 4).
All applicable national guidelines for the care and use of animals were followed, and the study was approved by the Institutional Committee of Bioethics for Animal Research (CIBICA) of the University of Valparaíso, Chile. Samples were remotely obtained using a hollow-tipped biopsy dart fired from a PaxArms .22 rifle that extracts a small sample of outer blubber, skin, and hair, with minimal disturbance of the animal. Date, sample site, and age/sex class of the individual were registered for each sampling event. Three age/sex classes were considered: (1) subadult males, (2) adult males, and (3) adult females, according to differences in size, body shape, coloration and, in the case of adult males, the presence of a developed mane 2 . Juveniles were not included in our study due to their small sample size. For this study we only used the outer blubber. Although the inner layer of the blubber (the section closest to the muscle) is known to be actively involved in the dietary metabolism reflecting the last weeks to months of feeding 88,89 , here, due to the sampling method utilised, we use only the outer layer (the section just beneath the skin), which has a more structural role 90 , is more stable and thought to reflect a longer-term diet than the inner layer (more than 2-3 months) 35,91 . Thus, dietary inferences made here correspond to a general diet, rather than represent a specific time period.  36 . Farmed salmon values (n = 81) correspond to Atlantic salmon, and were calculated from Blanchet et al. 83 , Megdal et al. 85 , and Aursand et al. 86 . Freshwater seal values (n = 34) were calculated from Grahl-Nielsen et al. 97   www.nature.com/scientificreports www.nature.com/scientificreports/ Laboratory procedures. Lipid extraction was conducted following a modified Folch et al. 92 method described in Budge et al. 35 . In brief, samples were left overnight in a solution containing 2:1 chloroform: methanol and 0.01% of butylated hydroxytoluene. Samples were then homogenised, washed in a salt solution, centrifuged to separate components, dried over anhydrous sodium sulphate and evaporated using nitrogen. Lipids obtained from the procedure were weighted and used for fatty acid transesterification.
We used an acidic transesterification method 35 , using Hilditch reagent, to prepare fatty acid methyl esters (FAMEs), which were then extracted into hexane and stored in vial tubes for gas chromatography analysis.
Gas chromatography analyses of FAMEs were conducted at the Mark Wainwright Analytical Centre within the University of New South Wales (Sydney, Australia). FAMEs were analysed in an Agilent 7890 A Series Gas Chromatograph System (Agilent Technologies, U.S.A.) equipped with a flame ionization detector, and using a flexible fused silica column DB-23 of 30 m length × 0.25 mm ID × 0.25 µm film (Agilent Technologies, U.S.A.). Operating parameters of the gas chromatograph have been described in Guerrero and Rogers 32 . FAMEs were identified using a range of standard mixtures (Nu Check Prep., Elysian, MN, USA) and concentrations were converted to percentage contributions of the total fatty acids. Data analysis. Due to noise contribution of fatty acids found in trace amounts 93 , only those fatty acids found in proportions higher than 0.5% were used for statistical analyses. Subadult and adult males were combined into a single group, due to their small sample sizes, making comparisons more numerically balanced. In order to assess separation of groups in multivariate space, we used nonmetric multidimensional scaling (MDS) analysis on untransformed fatty acid data. Permutational multivariate analysis of variance (PERMANOVA) based on Bray-Curtis dissimilarities was used to determine differences in fatty acid composition between groups. In order to account for the effect of sex on the differences in fatty acids between regions, we used a nested PERMANOVA, where sex was nested within each region. Using a two-way PERMANOVA, we tested for differences among colonies and sexes separately for each region, including sex, colony and an interaction term as independent variables. Fatty acids primarily influencing the separation among groups were identified using the similarity percentages routine (SIMPER). All these analyses were conducted using the R package "vegan" 94 .
Additionally, we tested for differences in dietary fatty acids typically used as trophic biomarkers, which included: total proportion of ω3 and ω6 fatty acids, and total proportion of isomers of C20:1 and C22:1. Prior to these statistical analyses, we applied the log-ratio transformation introduced by Aitchison 95 to our data. Although Budge et al. 35 recommended the use of C18:0 as the reference fatty acid for this transformation, here this compound varied substantially between sexes and regions (Table 1); therefore, we used C15:0 instead, since this fatty acid provides little dietary information, was reliably quantified, and did not contribute to differences between regions and/or sexes. Thus, we used Analysis of Variance (ANOVA) to evaluate differences in these fatty acid groups between regions, and sex within each region. The level of significance for all statistical tests was set at P < 0.05. All analyses were conducted using R Studio 96 .

Data availability
All data supporting the conclusions of this article are within the paper.