Dietary vs non-dietary fatty acid profiles of lake trout morphs from Lake Superior and Great Bear Lake: Are fish really what they eat?

Fatty acids are well-established biomarkers used to characterize trophic ecology, food-web linkages, and organismal niche for a diversity of taxa. Most often, fatty acids examined include only those previously identified as “dietary” or “extended dietary”. Consequently, non-dietary fatty acids are commonly discarded, representing approximately half of the fatty acids that can be extracted from fish muscle. In this study, we explored whether non-dietary fatty acids can provide worthwhile information by assessing their ability to discriminate intraspecific diversity within and between lakes. Non-dietary fatty acids delineated variation geographically (between lakes), between locations within a lake, and among morphs within a species. Non-dietary fatty acids were useful in delineating morphs of lake trout in both lakes and, for Lake Superior, performed better than dietary fatty acids. Physiological differences that arise from dynamic differences in energy processing can be adaptive and linked to habitat use (i.e., shallow vs. deep depths) by species morphs, and likely explains why non-dietary fatty acids can be a relevant tool to delineate intraspecific diversity in lake trout, which are known to inhabit a large range of habitats. The level of differences in non-dietary fatty acids were as pronounced among morphs within Great Bear Lake as observed in Lake Superior. Little is known what causes non-dietary-mediated differences in fatty acid composition, but our results showed that assessing non-dietary fatty acids can be useful in delineating intraspecific diversity and spatial variation.

Abstract: 23 Fatty acids are well-established biomarkers used to characterize trophic ecology, food-web 24 linkages, and organismal niche for a diversity of taxa. Most often, fatty acids examined include 25 only those previously identified as "dietary" or "extended dietary". Consequently, non-dietary 26 fatty acids are commonly discarded, representing approximately half of the fatty acids that can be 27 extracted from fish muscle. In this study, we explored whether non-dietary fatty acids can provide 28 worthwhile information by assessing their ability to discriminate intraspecific diversity within and 29 between lakes. Non-dietary fatty acids delineated variation geographically (between lakes), 30 between locations within a lake, and among morphs within a species. Non-dietary fatty acids were 31 useful in delineating morphs of lake trout in both lakes and, for Lake Superior, performed better 32 than dietary fatty acids. Physiological differences that arise from dynamic differences in energy 33 processing can be adaptive and linked to habitat use (i.e., shallow vs. deep depths) by species 34 morphs, and likely explains why non-dietary fatty acids can be a relevant tool to delineate 35 intraspecific diversity in lake trout, which are known to inhabit a large range of habitats. The level 36 of differences in non-dietary fatty acids were as pronounced among morphs within Great Bear 37 Lake as observed in Lake Superior. Little is known what causes non-dietary-mediated differences 38 in fatty acid composition, but our results showed that assessing non-dietary fatty acids can be 39 useful in delineating intraspecific diversity and spatial variation. Introduction: 44 Constraints on traditional methods to investigate diets of organisms within aquatic systems 45 have led to the use of biochemical tracers in contemporary investigations (Vinson & Budy, 2010). 46 Among these, fatty acids have gained popularity as both qualitative and quantitative trophic 47 markers that reflect foraging patterns and food-web dynamics (Iverson, 2009; Galloway et al.,  Fatty acids are the "building blocks" of lipids and represent the largest constituent of 54 neutral lipids (e.g., triacylglycerols and wax esters) and polar phospholipids (Iverson, 2009). The 55 array of fatty acids present in nature is exceptionally complex, routinely ~70 fatty acids can be 56 identified within an organism (Iverson, 2009;Budge, Iverson & Koopman, 2006). The utility of 57 fatty acid analyses to reflect foraging patterns and food web dynamics relies on the assumption 58 that lipids are broken down into their constituent fatty acids and incorporated relatively unchanged 59 into consumer tissues (Howell et al., 2003;Iverson et al., 2004;Iverson, 2009). The storage 60 patterns of fatty acids depend on the biochemical limitations of organisms to biosynthesize, modify 61 chain-length, and introduce double bonds in fatty acids (e.g., with complexity culminating in 62 vertebrates) (Iverson 2009). A critical analytical assumption is that storage characteristics of fatty 63 acids within lipids are not being degraded during digestion, i.e., that they are stored in tissues in 64 their basic form (Iverson, 2009). Fatty acids labelled as "dietary" or "extended dietary" tracers for analyses in ecological studies. Consequently, approximately half of the fatty acids extracted 67 from tissue samples are commonly omitted, with only those recognized to be typical markers of 68 diet being analyzed.

69
The exclusion of these "non-dietary" fatty acids from analyses is not misguided, as the 70 purpose of most studies is to accurately describe diet patterns, and this practice has resulted in 71 reliable information being produced across taxa (Iverson et al., 2004;Galloway et al., 2014;72 Iverson, 2009). However, it is unknown whether valid information for other research questions 73 may be lost when investigators discard a portion of the fatty acids in their samples and the potential 74 discriminatory data that could have resulted if analysed. Biological (e.g., phenotypic and genetic) 75 and environmental variables have been shown to affect lipid and fatty acid composition in fishes 76 (Olsen & Skjervold, 1995). For example, temperature (Olsen, 1999;Farkas et al., 1980), salinity 77 (Borlongan & Benitez, 1992), and light (Ota & Yamada, 1971) have induced variation in lipid 78 composition of fish tissue. Thus, fatty acids not labelled as dietary markers could be useful when 79 the aim of a study is to delineate or better understand intraspecific diversity. To investigate these 80 questions, we compared "non-dietary" fatty acids among morphs of lake trout (Salvelinus 81 namaycush) in Lake Superior to those among lake trout morphs in Great Bear Lake, as these 82 morphs represent important intraspecific diversity in these lakes.

83
Salmonids, such as lake trout, inhabit evolutionarily young ecosystems believed to be 84 10,000 to 15,000 years old, e.g., post-glacial lakes colonized from non-glaciated refugia. The 85 depauperate communities of post-glacial lakes are commonly characterized by reduced 86 competition and predation, which allows colonizers access to a relatively wide array of resources, 87 conditions that favour development of intraspecific diversity (Smith & Skulason, 1996;McPhail, 88 1993; Robinson & Wilson, 1994 Intraspecific diversity in lake trout has been mostly linked to differences in depth 98 distribution and, not surprisingly, is best known from large (> 500km 2 ), deep lakes, such as Lake   Great Bear Lake is the most northerly lake of its size (~31 000 km 2 ), has a maximum depth of 446 132 m, and is the fifteenth deepest freshwater lake in the world ( Fig. 1; Johnson, 1975). A UNESCO 133 biosphere reserve, Great Bear Lake is 250 km south of the Arctic Ocean and has characteristics 134 typical of an Arctic lake. The lake is ultra-oligotrophic and, despite its size, has a simple food web, 135 supporting only 15 fish species (Johnson, 1975;MacDonald et al., 2004). The lake and its biota 136 have remained relatively isolated and unexploited, and is one of the most pristine large lakes in 137 North America. Samples of lake trout were collected from multiple locations, Great Bear Lake has 138 five semi-isolated arms, but due to sample size, data were pooled across multiple sites (see 139 Chavarie et al., 2016b for details).

140
Great Bear Lake sustains a noteworthy example of lake trout divergence (Fig. 2). With its 141 intraspecific diversity independent of depth-based segregation, the lake also presents an unusual  The humper lake trout has a small head, short snout, short maxillae, large eyes, and short and  overnight extraction (at -20) in 2:1 chloroform:methanol containing 0.01% BHT (v/v/w) (Folch,179 Lees & Sloane-Stanley, 1957), samples were filtered through Whatman Grade 1 Qualitative filter 180 paper and the filter paper sample was rinsed twice with 2 mL of 2:1 chloroform:methanol. Sample 181 extract was collected in a test tube and 7 mL 0.88 NaCl solution were added because NaCl 182 encourages fatty acids to move into the organic (chloroform) layer. The aqueous layer was 183 discarded after which the chloroform was dried with sodium sulfate prior to total lipid 184 determination. The extracted lipid was used to prepare fatty acid methyl esters (FAME) by 185 transesterification with Hilditch reagent (0.5 N H2SO4 in methanol) (Morrison & Smith, 1964).  Thirty-eight dietary and 24 non-dietary fatty acids were found to be shared between the two lakes, 210 and these were selected for further analyses (Table A1 and A2).  To test for differences in fatty acid composition among morphs within each lake (dietary,

235
Combined lakes analyses 236 The first two axes of non-dietary fatty acids PCA explained 42.4 % of the variation among 237 lake trout from Great Bear Lake and Lake Superior (Fig. 3). Lake trout from Great Bear Lake 238 clustered together with 37 lake trout overlapped with those from Lake Superior, whereas those Great Bear Lake and Lake Superior. Lake trout from Great Bear Lake grouped together, with 32 244 individuals overlapped with Lake Superior lake trout, whereas lake trout from Stannard Rock 245 totally overlapped with Superior Shoal individuals (Fig. 3). Separation between the two lakes 246 appeared to be driven by fatty acids associated with pelagic habitat (14:0, 20:1n-9, 20:1n-7, 20:1n-247 11 and 22:1n-9; toward Lake Superior) versus one fatty acid associated with cannibalism or/and 248 carnivory (20:5n3; toward Great Bear Lake) (Appendix , Table 1). Finally, the first two axes of the 249 PCA based on all fatty acids combined explained 39.0% of the variation among lake trout morphs 250 from Great Bear Lake and Lake Superior. Lake trout from Great Bear Lake clustered together but 251 with 15 lake trout overlapped with those from Lake Superior. Lake trout from Stannard Rock and 252 Superior Shoal in Lake Superior overlapped completely.  (Table 1). The first two axes of the linear discriminant 260 analysis explained 55.7 % and 27.8% of the variation, and 57.9% of all individuals were correctly 261 classified to morph based on non-dietary fatty acids (Fig. 4). Dietary fatty acids also differed 262 among the four lake trout morphs from Great Bear Lake (one-way PERMANOVA, F3,122 = 2.95, 263 P < 0.01), and most pairs of morphs differed significantly from one another (all P < 0.01 except 264 for morphs 1 vs. 3). The ten most discriminating dietary fatty acids from SIMPER explained 265 ~48.8% of the dissimilarity among groups ( Table 2). The first two axes of the linear discriminant 266 analysis explained 46.7% and 39.0% of the variation, and 68.3% of all individuals were correctly 267 classified to morph based on dietary fatty acids (Fig. 4). Finally, when the 10 dietary and 10 non-268 dietary most discriminating fatty acids were considered together, morphs differed from one another 269 (PERMANOVA, F3,122 = 2.8, P < 0.01). All pairs of morphs differed (P < 0.01) except for morphs 270 1 and 3. The first two axes of the linear discriminant analysis explained 63.6% and 24.9% of the 271 variation, and 54.8% of all individuals were correctly classified (Fig. 4). were correctly classified to morph based on dietary fatty acids (Fig. 4). Finally, the three morphs 285 also differed when the 10 most discriminating non-dietary and dietary fatty acids were considered 286 together (one-way PERMANOVA, F2,87 = 2.1, P < 0.02), and leans differed significantly from 287 siscowets (P < 0.01). The first two axes of the linear discriminant analysis explained 73.3% and 288 26.8% of the variation, respectively, and 48.9% of all individuals were correctly classified to 289 morph based on their dietary fatty acid profile (Fig. 4). four morphs (one-way PERMANOVA, F3,116 = 2.5, P < 0.01), and leans differed significantly 293 from redfins (P < 0.01) and were marginally significantly different between redfin and siscowet 294 and between redfin and humper (P < 0.06). The ten most discriminating non-dietary fatty acids 295 from SIMPER explained ~69.9% of the dissimilarity among groups (Table 1). The first two axes 296 of the linear discriminant analysis explained 64.3% and 25.6% of the variation, and 45.0% 297 individuals were correctly classified to morph using non-dietary fatty acids (Fig. 4). Based on 298 dietary fatty acid composition, we found no differences among the four lake trout morphs from 299 Superior Shoal (one-way PERMANOVA, F3,116 = 0.8, P = 0.3), similarly to what was found with 300 fish from Stannard Rock. The ten most discriminating fatty acids from SIMPER explained ~55.8 301 % of the dissimilarity among groups ( Table 2). The first two axes of the linear discriminant 302 analysis explained 38.6% and 35.6% of the variation, respectively, and 31.7% individuals were 303 correctly classified to morph based on their dietary fatty acid profile (Fig. 4). In contrast to the 304 results for Great Bear Lake and Stannard Rock, when the 10 non-dietary and 10 dietary most 305 discriminating fatty acids were combined for lake trout from Superior Shoal, morphs did not differ 306 significantly from one another (one-way PERMANOVA, F3,116 = 1.3, P = 0.16). The first two 307 axes of the linear discriminant analysis explained 72.9% and 19.7% of the variation, respectively, 308 and 42.5% of all individuals were correctly classified (Fig. 4). acids showed variation geographically (between lakes), between locations within a lake (i.e., 316 Stannard Rock versus Superior Shoal), and among morphs within a species. Although some 317 overlap was found, our results showed that when investigating intraspecific diversity, non-dietary 318 fatty acids can be a useful tool to delineate groups, and that sometimes, such as at sites in Lake 319 Superior, were more discriminatory than dietary fatty acids. Accordingly, our results suggested  to changes in the prey base also deserves more consideration.

361
The concept that dynamics of energy processing and storage are adaptive along a gradient 362 that is at least partially driven by depth does not apply to the intraspecific diversity of lake trout in 363 Great Bear Lake. This diversity is limited to shallow-water habitat, and appears to be independent 364 of major habitat and resource partitioning (Chavarie et al., 2016a; Chavarie et al., 2018). The 365 similarity of results between Great Bear Lake and Lake Superior is thus perplexing, as we were 366 expected greater differences in non-dietary fatty acids among morphs in Lake Superior than in 367 Great Bear Lake, due to known buoyancy variation associated with a depth gradient in Lake 368 Superior. If lake trout morphs from Great Bear Lake are also under selection (Harris et al., 2014), 369 differences in energy processing and storage may be as pronounced as those that have been 370 observed in Lake Superior. Another question raised by our results is the extent to which morphs 371 are independent of major habitat or resource partitioning (e.g., same question is pertinent for Lake 372 Superior), especially because dietary fatty acids in Great Bear Lake were slightly better at 373 delineating intraspecific groups than non-dietary fatty acids.

374
Ecological differences in allopatry are often found to be more pronounced than those in 375 sympatry, due to disparate environments and isolation (Rundle & Nosil, 2005;Yoder et al., 2010;376 Heggenes, 2002;Fraser et al., 2011). In this study, differences between lakes in both dietary and 377 non-dietary fatty acids appear to be greater than variation among morphs within a lake. For dietary 378 fatty acids, differences between the two lakes appeared to be due to fatty acids associated with a 379 pelagic environment, such as C20 and C22 monounsaturates, that can be used as biomarkers of Great Bear Lake and Lake Superior, considerable overlap occurred among dietary (e.g., five out 390 of 10) and non-dietary (e.g., 7 out of 10) fatty acids that were important in identifying intraspecific 391 diversity in both lakes. The greater number of shared non-dietary fatty acids discriminating lake 392 trout intraspecific diversity in the two lakes supports the idea of similar physiological differences 393 (e.g., energy processing and storage) among morphs from both lakes.